node-autotask 0.86.0 → 0.87.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (38) hide show
  1. package/CHANGELOG.md +7 -0
  2. package/MCP_TOOL_CATALOG.json +2 -2
  3. package/capabilities.json +175 -175
  4. package/dist/capabilities-data.js +174 -174
  5. package/dist/capabilities.cjs +4 -4
  6. package/dist/capabilities.js +1 -1
  7. package/dist/{chunk-YX6SSS4I.cjs → chunk-CUZJJCI5.cjs} +420 -219
  8. package/dist/{chunk-KYFJOIG4.js → chunk-EXNDYJPG.js} +3 -3
  9. package/dist/{chunk-DVD7RZIB.js → chunk-G57Q7OAC.js} +2 -2
  10. package/dist/{chunk-ELZTECDS.js → chunk-GAFG7RHG.js} +415 -214
  11. package/dist/{chunk-23LTWBKP.cjs → chunk-HUA4WB4N.cjs} +30 -30
  12. package/dist/{chunk-AZ54RMQR.js → chunk-NXSQXXNK.js} +2 -2
  13. package/dist/{chunk-SZ33SCCI.cjs → chunk-RU4ECNQN.cjs} +200 -200
  14. package/dist/{chunk-EYRQUMA2.cjs → chunk-YVNGUGG6.cjs} +3 -3
  15. package/dist/{index-B3zO78nP.d.ts → index-CsL5gGJo.d.ts} +8 -8
  16. package/dist/{index-CIgmz8qI.d.cts → index-Cxfr_DOD.d.cts} +8 -2
  17. package/dist/{index-CIgmz8qI.d.ts → index-Cxfr_DOD.d.ts} +8 -2
  18. package/dist/{index-C2p8D2mZ.d.cts → index-DGiua9uK.d.cts} +8 -8
  19. package/dist/index.cjs +84 -84
  20. package/dist/index.d.cts +3 -3
  21. package/dist/index.d.ts +3 -3
  22. package/dist/index.js +5 -5
  23. package/dist/operations/index.cjs +51 -51
  24. package/dist/operations/index.d.cts +4 -4
  25. package/dist/operations/index.d.ts +4 -4
  26. package/dist/operations/index.js +2 -2
  27. package/dist/pool.cjs +12 -12
  28. package/dist/pool.d.cts +1 -1
  29. package/dist/pool.d.ts +1 -1
  30. package/dist/pool.js +4 -4
  31. package/dist/resources/index.cjs +27 -27
  32. package/dist/resources/index.d.cts +2 -2
  33. package/dist/resources/index.d.ts +2 -2
  34. package/dist/resources/index.js +3 -3
  35. package/dist/types/index.d.cts +1 -1
  36. package/dist/types/index.d.ts +1 -1
  37. package/docs/VENDOR-QUIRKS.md +34 -17
  38. package/package.json +1 -1
package/capabilities.json CHANGED
@@ -1,9 +1,9 @@
1
1
  {
2
- "generatedAt": "2026-11-20T10:01:10.506Z",
2
+ "generatedAt": "2026-11-20T11:02:13.927Z",
3
3
  "groups": {
4
4
  "B": {
5
- "planHash": "929593354ffa904a038cb7cbbf7b91ffbb05e9f06078181e9f225505f3463a88",
6
- "builtAt": "2026-11-20T10:01:10.493Z",
5
+ "planHash": "c97deac4ac7c3ca9ce4cea10ac6a50bb19b6517993d9657c52397964e1fcf715",
6
+ "builtAt": "2026-11-20T11:02:13.903Z",
7
7
  "records": [
8
8
  {
9
9
  "id": "appointments.count",
@@ -396,7 +396,7 @@
396
396
  "params": {
397
397
  "type": "object",
398
398
  "required": false,
399
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
399
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
400
400
  "properties": [
401
401
  {
402
402
  "name": "filter",
@@ -955,7 +955,7 @@
955
955
  "params": {
956
956
  "type": "object",
957
957
  "required": false,
958
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
958
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
959
959
  "properties": [
960
960
  {
961
961
  "name": "filter",
@@ -1384,7 +1384,7 @@
1384
1384
  "params": {
1385
1385
  "type": "object",
1386
1386
  "required": false,
1387
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
1387
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
1388
1388
  "properties": [
1389
1389
  {
1390
1390
  "name": "filter",
@@ -1773,7 +1773,7 @@
1773
1773
  "params": {
1774
1774
  "type": "object",
1775
1775
  "required": false,
1776
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
1776
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
1777
1777
  "properties": [
1778
1778
  {
1779
1779
  "name": "filter",
@@ -2369,7 +2369,7 @@
2369
2369
  "params": {
2370
2370
  "type": "object",
2371
2371
  "required": false,
2372
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
2372
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
2373
2373
  "properties": [
2374
2374
  {
2375
2375
  "name": "filter",
@@ -2921,7 +2921,7 @@
2921
2921
  "params": {
2922
2922
  "type": "object",
2923
2923
  "required": false,
2924
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
2924
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
2925
2925
  "properties": [
2926
2926
  {
2927
2927
  "name": "filter",
@@ -3619,7 +3619,7 @@
3619
3619
  "params": {
3620
3620
  "type": "object",
3621
3621
  "required": false,
3622
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
3622
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
3623
3623
  "properties": [
3624
3624
  {
3625
3625
  "name": "filter",
@@ -4132,7 +4132,7 @@
4132
4132
  "params": {
4133
4133
  "type": "object",
4134
4134
  "required": false,
4135
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
4135
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
4136
4136
  "properties": [
4137
4137
  {
4138
4138
  "name": "filter",
@@ -4710,7 +4710,7 @@
4710
4710
  "params": {
4711
4711
  "type": "object",
4712
4712
  "required": false,
4713
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
4713
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
4714
4714
  "properties": [
4715
4715
  {
4716
4716
  "name": "filter",
@@ -5387,7 +5387,7 @@
5387
5387
  "params": {
5388
5388
  "type": "object",
5389
5389
  "required": false,
5390
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
5390
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
5391
5391
  "properties": [
5392
5392
  {
5393
5393
  "name": "filter",
@@ -5993,7 +5993,7 @@
5993
5993
  "params": {
5994
5994
  "type": "object",
5995
5995
  "required": false,
5996
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
5996
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
5997
5997
  "properties": [
5998
5998
  {
5999
5999
  "name": "filter",
@@ -6649,7 +6649,7 @@
6649
6649
  "params": {
6650
6650
  "type": "object",
6651
6651
  "required": false,
6652
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
6652
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
6653
6653
  "properties": [
6654
6654
  {
6655
6655
  "name": "filter",
@@ -7194,7 +7194,7 @@
7194
7194
  "params": {
7195
7195
  "type": "object",
7196
7196
  "required": false,
7197
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
7197
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
7198
7198
  "properties": [
7199
7199
  {
7200
7200
  "name": "filter",
@@ -7867,7 +7867,7 @@
7867
7867
  "params": {
7868
7868
  "type": "object",
7869
7869
  "required": false,
7870
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
7870
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
7871
7871
  "properties": [
7872
7872
  {
7873
7873
  "name": "filter",
@@ -8406,7 +8406,7 @@
8406
8406
  "params": {
8407
8407
  "type": "object",
8408
8408
  "required": false,
8409
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
8409
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
8410
8410
  "properties": [
8411
8411
  {
8412
8412
  "name": "filter",
@@ -8958,7 +8958,7 @@
8958
8958
  "params": {
8959
8959
  "type": "object",
8960
8960
  "required": false,
8961
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
8961
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
8962
8962
  "properties": [
8963
8963
  {
8964
8964
  "name": "filter",
@@ -9539,7 +9539,7 @@
9539
9539
  "params": {
9540
9540
  "type": "object",
9541
9541
  "required": false,
9542
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
9542
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
9543
9543
  "properties": [
9544
9544
  {
9545
9545
  "name": "filter",
@@ -10120,7 +10120,7 @@
10120
10120
  "params": {
10121
10121
  "type": "object",
10122
10122
  "required": false,
10123
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
10123
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
10124
10124
  "properties": [
10125
10125
  {
10126
10126
  "name": "filter",
@@ -10685,7 +10685,7 @@
10685
10685
  "params": {
10686
10686
  "type": "object",
10687
10687
  "required": false,
10688
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
10688
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
10689
10689
  "properties": [
10690
10690
  {
10691
10691
  "name": "filter",
@@ -11049,7 +11049,7 @@
11049
11049
  "params": {
11050
11050
  "type": "object",
11051
11051
  "required": false,
11052
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
11052
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
11053
11053
  "properties": [
11054
11054
  {
11055
11055
  "name": "filter",
@@ -11413,7 +11413,7 @@
11413
11413
  "params": {
11414
11414
  "type": "object",
11415
11415
  "required": false,
11416
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
11416
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
11417
11417
  "properties": [
11418
11418
  {
11419
11419
  "name": "filter",
@@ -11851,7 +11851,7 @@
11851
11851
  "params": {
11852
11852
  "type": "object",
11853
11853
  "required": false,
11854
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
11854
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
11855
11855
  "properties": [
11856
11856
  {
11857
11857
  "name": "filter",
@@ -12484,7 +12484,7 @@
12484
12484
  "params": {
12485
12485
  "type": "object",
12486
12486
  "required": false,
12487
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
12487
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
12488
12488
  "properties": [
12489
12489
  {
12490
12490
  "name": "filter",
@@ -13273,7 +13273,7 @@
13273
13273
  "params": {
13274
13274
  "type": "object",
13275
13275
  "required": false,
13276
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
13276
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
13277
13277
  "properties": [
13278
13278
  {
13279
13279
  "name": "filter",
@@ -13916,7 +13916,7 @@
13916
13916
  "params": {
13917
13917
  "type": "object",
13918
13918
  "required": false,
13919
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
13919
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
13920
13920
  "properties": [
13921
13921
  {
13922
13922
  "name": "filter",
@@ -14591,7 +14591,7 @@
14591
14591
  "params": {
14592
14592
  "type": "object",
14593
14593
  "required": false,
14594
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
14594
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
14595
14595
  "properties": [
14596
14596
  {
14597
14597
  "name": "filter",
@@ -15150,7 +15150,7 @@
15150
15150
  "params": {
15151
15151
  "type": "object",
15152
15152
  "required": false,
15153
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
15153
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
15154
15154
  "properties": [
15155
15155
  {
15156
15156
  "name": "filter",
@@ -15919,7 +15919,7 @@
15919
15919
  "params": {
15920
15920
  "type": "object",
15921
15921
  "required": false,
15922
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
15922
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
15923
15923
  "properties": [
15924
15924
  {
15925
15925
  "name": "filter",
@@ -16548,7 +16548,7 @@
16548
16548
  "params": {
16549
16549
  "type": "object",
16550
16550
  "required": false,
16551
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
16551
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
16552
16552
  "properties": [
16553
16553
  {
16554
16554
  "name": "filter",
@@ -17163,7 +17163,7 @@
17163
17163
  "params": {
17164
17164
  "type": "object",
17165
17165
  "required": false,
17166
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
17166
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
17167
17167
  "properties": [
17168
17168
  {
17169
17169
  "name": "filter",
@@ -17858,7 +17858,7 @@
17858
17858
  "params": {
17859
17859
  "type": "object",
17860
17860
  "required": false,
17861
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
17861
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
17862
17862
  "properties": [
17863
17863
  {
17864
17864
  "name": "filter",
@@ -18389,7 +18389,7 @@
18389
18389
  "params": {
18390
18390
  "type": "object",
18391
18391
  "required": false,
18392
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
18392
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
18393
18393
  "properties": [
18394
18394
  {
18395
18395
  "name": "filter",
@@ -18964,7 +18964,7 @@
18964
18964
  "params": {
18965
18965
  "type": "object",
18966
18966
  "required": false,
18967
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
18967
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
18968
18968
  "properties": [
18969
18969
  {
18970
18970
  "name": "filter",
@@ -19455,7 +19455,7 @@
19455
19455
  "params": {
19456
19456
  "type": "object",
19457
19457
  "required": false,
19458
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
19458
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
19459
19459
  "properties": [
19460
19460
  {
19461
19461
  "name": "filter",
@@ -19908,7 +19908,7 @@
19908
19908
  "params": {
19909
19909
  "type": "object",
19910
19910
  "required": false,
19911
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
19911
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
19912
19912
  "properties": [
19913
19913
  {
19914
19914
  "name": "filter",
@@ -20522,7 +20522,7 @@
20522
20522
  "params": {
20523
20523
  "type": "object",
20524
20524
  "required": false,
20525
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
20525
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
20526
20526
  "properties": [
20527
20527
  {
20528
20528
  "name": "filter",
@@ -21014,7 +21014,7 @@
21014
21014
  "params": {
21015
21015
  "type": "object",
21016
21016
  "required": false,
21017
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
21017
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
21018
21018
  "properties": [
21019
21019
  {
21020
21020
  "name": "filter",
@@ -21514,7 +21514,7 @@
21514
21514
  "params": {
21515
21515
  "type": "object",
21516
21516
  "required": false,
21517
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
21517
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
21518
21518
  "properties": [
21519
21519
  {
21520
21520
  "name": "filter",
@@ -22001,7 +22001,7 @@
22001
22001
  "params": {
22002
22002
  "type": "object",
22003
22003
  "required": false,
22004
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
22004
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
22005
22005
  "properties": [
22006
22006
  {
22007
22007
  "name": "filter",
@@ -22701,7 +22701,7 @@
22701
22701
  "params": {
22702
22702
  "type": "object",
22703
22703
  "required": false,
22704
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
22704
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
22705
22705
  "properties": [
22706
22706
  {
22707
22707
  "name": "filter",
@@ -23374,7 +23374,7 @@
23374
23374
  "params": {
23375
23375
  "type": "object",
23376
23376
  "required": false,
23377
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
23377
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
23378
23378
  "properties": [
23379
23379
  {
23380
23380
  "name": "filter",
@@ -23928,7 +23928,7 @@
23928
23928
  "params": {
23929
23929
  "type": "object",
23930
23930
  "required": false,
23931
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
23931
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
23932
23932
  "properties": [
23933
23933
  {
23934
23934
  "name": "filter",
@@ -24432,7 +24432,7 @@
24432
24432
  "params": {
24433
24433
  "type": "object",
24434
24434
  "required": false,
24435
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
24435
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
24436
24436
  "properties": [
24437
24437
  {
24438
24438
  "name": "filter",
@@ -25013,7 +25013,7 @@
25013
25013
  "params": {
25014
25014
  "type": "object",
25015
25015
  "required": false,
25016
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
25016
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
25017
25017
  "properties": [
25018
25018
  {
25019
25019
  "name": "filter",
@@ -25530,7 +25530,7 @@
25530
25530
  "params": {
25531
25531
  "type": "object",
25532
25532
  "required": false,
25533
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
25533
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
25534
25534
  "properties": [
25535
25535
  {
25536
25536
  "name": "filter",
@@ -26144,7 +26144,7 @@
26144
26144
  "params": {
26145
26145
  "type": "object",
26146
26146
  "required": false,
26147
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
26147
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
26148
26148
  "properties": [
26149
26149
  {
26150
26150
  "name": "filter",
@@ -26570,7 +26570,7 @@
26570
26570
  "params": {
26571
26571
  "type": "object",
26572
26572
  "required": false,
26573
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
26573
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
26574
26574
  "properties": [
26575
26575
  {
26576
26576
  "name": "filter",
@@ -27059,7 +27059,7 @@
27059
27059
  "params": {
27060
27060
  "type": "object",
27061
27061
  "required": false,
27062
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
27062
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
27063
27063
  "properties": [
27064
27064
  {
27065
27065
  "name": "filter",
@@ -27502,7 +27502,7 @@
27502
27502
  "params": {
27503
27503
  "type": "object",
27504
27504
  "required": false,
27505
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
27505
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
27506
27506
  "properties": [
27507
27507
  {
27508
27508
  "name": "filter",
@@ -28304,7 +28304,7 @@
28304
28304
  "params": {
28305
28305
  "type": "object",
28306
28306
  "required": false,
28307
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
28307
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
28308
28308
  "properties": [
28309
28309
  {
28310
28310
  "name": "filter",
@@ -29175,7 +29175,7 @@
29175
29175
  "params": {
29176
29176
  "type": "object",
29177
29177
  "required": false,
29178
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
29178
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
29179
29179
  "properties": [
29180
29180
  {
29181
29181
  "name": "filter",
@@ -30044,7 +30044,7 @@
30044
30044
  "params": {
30045
30045
  "type": "object",
30046
30046
  "required": false,
30047
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
30047
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
30048
30048
  "properties": [
30049
30049
  {
30050
30050
  "name": "filter",
@@ -30719,7 +30719,7 @@
30719
30719
  "params": {
30720
30720
  "type": "object",
30721
30721
  "required": false,
30722
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
30722
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
30723
30723
  "properties": [
30724
30724
  {
30725
30725
  "name": "filter",
@@ -31221,7 +31221,7 @@
31221
31221
  "params": {
31222
31222
  "type": "object",
31223
31223
  "required": false,
31224
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
31224
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
31225
31225
  "properties": [
31226
31226
  {
31227
31227
  "name": "filter",
@@ -31806,7 +31806,7 @@
31806
31806
  "params": {
31807
31807
  "type": "object",
31808
31808
  "required": false,
31809
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
31809
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
31810
31810
  "properties": [
31811
31811
  {
31812
31812
  "name": "filter",
@@ -32343,7 +32343,7 @@
32343
32343
  "params": {
32344
32344
  "type": "object",
32345
32345
  "required": false,
32346
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
32346
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
32347
32347
  "properties": [
32348
32348
  {
32349
32349
  "name": "filter",
@@ -32880,7 +32880,7 @@
32880
32880
  "params": {
32881
32881
  "type": "object",
32882
32882
  "required": false,
32883
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
32883
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
32884
32884
  "properties": [
32885
32885
  {
32886
32886
  "name": "filter",
@@ -33417,7 +33417,7 @@
33417
33417
  "params": {
33418
33418
  "type": "object",
33419
33419
  "required": false,
33420
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
33420
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
33421
33421
  "properties": [
33422
33422
  {
33423
33423
  "name": "filter",
@@ -33765,7 +33765,7 @@
33765
33765
  "opts": {
33766
33766
  "type": "object",
33767
33767
  "required": false,
33768
- "description": "CompanyFullProfileOptions — { contactLimit?: number (default 50), noteLimit?: number (default 50), ticketLimit?: number (default 50), projectLimit?: number (default 50), locationLimit?: number (default 50, 1-500), fields?: PartFields (per-part field projection: keys locations | contacts | notes | projects | openTickets, each a non-empty list of exact vendor field names — a listed part serves ONLY those fields (IncludeFields on the wire), an absent part keeps the full model; an unknown part key or field name is refused pre-request, zero wire; with names: true the name siblings appear only for the id fields kept), names?: boolean (default false — opt-in name/label siblings + the top-level `names` section), maxNameReads?: number (default 20 — cold wire name-read budget, lowerable never raised), excludeStatusIds?: number[] (explicit closed ticket-status ids for openTickets, overriding the resolved picklist), allowAssumed?: boolean (default true; false throws STATUS_SEMANTICS_UNRESOLVED instead of using the assumed closed id 5) }.",
33768
+ "description": "CompanyFullProfileOptions — { contactLimit?: number (default 50), noteLimit?: number (default 50), ticketLimit?: number (default 50), projectLimit?: number (default 50), locationLimit?: number (default 50, 1-500), fields?: PartFields (per-part field projection: keys locations | contacts | notes | projects | openTickets, each a non-empty list of exact vendor field names — a listed part serves ONLY those fields (IncludeFields on the wire), an absent part keeps the full model; an unknown part key or field name is refused pre-request, zero wire; with names: true the name siblings appear only for the id fields kept), names?: boolean (default false — opt-in name/label siblings + the top-level `names` section), maxNameReads?: number (default 50 — cold wire name-read budget, lowerable never raised), excludeStatusIds?: number[] (explicit closed ticket-status ids for openTickets, overriding the resolved picklist), allowAssumed?: boolean (default true; false throws STATUS_SEMANTICS_UNRESOLVED instead of using the assumed closed id 5) }.",
33769
33769
  "properties": [
33770
33770
  {
33771
33771
  "name": "contactLimit",
@@ -33812,7 +33812,7 @@
33812
33812
  "name": "maxNameReads",
33813
33813
  "type": "number",
33814
33814
  "optional": true,
33815
- "default": 20
33815
+ "default": 50
33816
33816
  },
33817
33817
  {
33818
33818
  "name": "excludeStatusIds",
@@ -34081,7 +34081,7 @@
34081
34081
  "params": {
34082
34082
  "type": "object",
34083
34083
  "required": false,
34084
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
34084
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
34085
34085
  "properties": [
34086
34086
  {
34087
34087
  "name": "filter",
@@ -34584,7 +34584,7 @@
34584
34584
  "params": {
34585
34585
  "type": "object",
34586
34586
  "required": false,
34587
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
34587
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
34588
34588
  "properties": [
34589
34589
  {
34590
34590
  "name": "filter",
@@ -35223,7 +35223,7 @@
35223
35223
  "params": {
35224
35224
  "type": "object",
35225
35225
  "required": false,
35226
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
35226
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
35227
35227
  "properties": [
35228
35228
  {
35229
35229
  "name": "filter",
@@ -35892,7 +35892,7 @@
35892
35892
  "params": {
35893
35893
  "type": "object",
35894
35894
  "required": false,
35895
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
35895
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
35896
35896
  "properties": [
35897
35897
  {
35898
35898
  "name": "filter",
@@ -36392,7 +36392,7 @@
36392
36392
  "params": {
36393
36393
  "type": "object",
36394
36394
  "required": false,
36395
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
36395
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
36396
36396
  "properties": [
36397
36397
  {
36398
36398
  "name": "filter",
@@ -36815,7 +36815,7 @@
36815
36815
  "params": {
36816
36816
  "type": "object",
36817
36817
  "required": false,
36818
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
36818
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
36819
36819
  "properties": [
36820
36820
  {
36821
36821
  "name": "filter",
@@ -37263,7 +37263,7 @@
37263
37263
  "params": {
37264
37264
  "type": "object",
37265
37265
  "required": false,
37266
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
37266
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
37267
37267
  "properties": [
37268
37268
  {
37269
37269
  "name": "filter",
@@ -37903,7 +37903,7 @@
37903
37903
  "params": {
37904
37904
  "type": "object",
37905
37905
  "required": false,
37906
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
37906
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
37907
37907
  "properties": [
37908
37908
  {
37909
37909
  "name": "filter",
@@ -38454,7 +38454,7 @@
38454
38454
  "params": {
38455
38455
  "type": "object",
38456
38456
  "required": false,
38457
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
38457
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
38458
38458
  "properties": [
38459
38459
  {
38460
38460
  "name": "filter",
@@ -39065,7 +39065,7 @@
39065
39065
  "params": {
39066
39066
  "type": "object",
39067
39067
  "required": false,
39068
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
39068
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
39069
39069
  "properties": [
39070
39070
  {
39071
39071
  "name": "filter",
@@ -39625,7 +39625,7 @@
39625
39625
  "params": {
39626
39626
  "type": "object",
39627
39627
  "required": false,
39628
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
39628
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
39629
39629
  "properties": [
39630
39630
  {
39631
39631
  "name": "filter",
@@ -40107,7 +40107,7 @@
40107
40107
  "params": {
40108
40108
  "type": "object",
40109
40109
  "required": false,
40110
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
40110
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
40111
40111
  "properties": [
40112
40112
  {
40113
40113
  "name": "filter",
@@ -40698,7 +40698,7 @@
40698
40698
  "params": {
40699
40699
  "type": "object",
40700
40700
  "required": false,
40701
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
40701
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
40702
40702
  "properties": [
40703
40703
  {
40704
40704
  "name": "filter",
@@ -41151,7 +41151,7 @@
41151
41151
  "params": {
41152
41152
  "type": "object",
41153
41153
  "required": false,
41154
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
41154
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
41155
41155
  "properties": [
41156
41156
  {
41157
41157
  "name": "filter",
@@ -41774,7 +41774,7 @@
41774
41774
  "params": {
41775
41775
  "type": "object",
41776
41776
  "required": false,
41777
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
41777
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
41778
41778
  "properties": [
41779
41779
  {
41780
41780
  "name": "filter",
@@ -42338,7 +42338,7 @@
42338
42338
  "params": {
42339
42339
  "type": "object",
42340
42340
  "required": false,
42341
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
42341
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
42342
42342
  "properties": [
42343
42343
  {
42344
42344
  "name": "filter",
@@ -43627,7 +43627,7 @@
43627
43627
  "params": {
43628
43628
  "type": "object",
43629
43629
  "required": false,
43630
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
43630
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
43631
43631
  "properties": [
43632
43632
  {
43633
43633
  "name": "filter",
@@ -44578,7 +44578,7 @@
44578
44578
  "params": {
44579
44579
  "type": "object",
44580
44580
  "required": false,
44581
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
44581
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
44582
44582
  "properties": [
44583
44583
  {
44584
44584
  "name": "filter",
@@ -45146,7 +45146,7 @@
45146
45146
  "params": {
45147
45147
  "type": "object",
45148
45148
  "required": false,
45149
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
45149
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
45150
45150
  "properties": [
45151
45151
  {
45152
45152
  "name": "filter",
@@ -45789,7 +45789,7 @@
45789
45789
  "params": {
45790
45790
  "type": "object",
45791
45791
  "required": false,
45792
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
45792
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
45793
45793
  "properties": [
45794
45794
  {
45795
45795
  "name": "filter",
@@ -46307,7 +46307,7 @@
46307
46307
  "params": {
46308
46308
  "type": "object",
46309
46309
  "required": false,
46310
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
46310
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
46311
46311
  "properties": [
46312
46312
  {
46313
46313
  "name": "filter",
@@ -46603,8 +46603,8 @@
46603
46603
  ]
46604
46604
  },
46605
46605
  "A": {
46606
- "planHash": "929593354ffa904a038cb7cbbf7b91ffbb05e9f06078181e9f225505f3463a88",
46607
- "builtAt": "2026-11-20T10:01:10.493Z",
46606
+ "planHash": "c97deac4ac7c3ca9ce4cea10ac6a50bb19b6517993d9657c52397964e1fcf715",
46607
+ "builtAt": "2026-11-20T11:02:13.904Z",
46608
46608
  "records": [
46609
46609
  {
46610
46610
  "id": "attachment_info.count",
@@ -46885,7 +46885,7 @@
46885
46885
  "params": {
46886
46886
  "type": "object",
46887
46887
  "required": false,
46888
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
46888
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
46889
46889
  "properties": [
46890
46890
  {
46891
46891
  "name": "filter",
@@ -47176,7 +47176,7 @@
47176
47176
  "opts": {
47177
47177
  "type": "object",
47178
47178
  "required": false,
47179
- "description": "CompanyFullProfileOptions — { contactLimit?: number (default 50), noteLimit?: number (default 50), ticketLimit?: number (default 50), projectLimit?: number (default 50), locationLimit?: number (default 50, 1-500), fields?: PartFields (per-part field projection: keys locations | contacts | notes | projects | openTickets, each a non-empty list of exact vendor field names — a listed part serves ONLY those fields (IncludeFields on the wire), an absent part keeps the full model; an unknown part key or field name is refused pre-request, zero wire; with names: true the name siblings appear only for the id fields kept), names?: boolean (default false — opt-in name/label siblings + the top-level `names` section), maxNameReads?: number (default 20 — cold wire name-read budget, lowerable never raised), excludeStatusIds?: number[] (explicit closed ticket-status ids for openTickets, overriding the resolved picklist), allowAssumed?: boolean (default true; false throws STATUS_SEMANTICS_UNRESOLVED instead of using the assumed closed id 5) }.",
47179
+ "description": "CompanyFullProfileOptions — { contactLimit?: number (default 50), noteLimit?: number (default 50), ticketLimit?: number (default 50), projectLimit?: number (default 50), locationLimit?: number (default 50, 1-500), fields?: PartFields (per-part field projection: keys locations | contacts | notes | projects | openTickets, each a non-empty list of exact vendor field names — a listed part serves ONLY those fields (IncludeFields on the wire), an absent part keeps the full model; an unknown part key or field name is refused pre-request, zero wire; with names: true the name siblings appear only for the id fields kept), names?: boolean (default false — opt-in name/label siblings + the top-level `names` section), maxNameReads?: number (default 50 — cold wire name-read budget, lowerable never raised), excludeStatusIds?: number[] (explicit closed ticket-status ids for openTickets, overriding the resolved picklist), allowAssumed?: boolean (default true; false throws STATUS_SEMANTICS_UNRESOLVED instead of using the assumed closed id 5) }.",
47180
47180
  "properties": [
47181
47181
  {
47182
47182
  "name": "contactLimit",
@@ -47223,7 +47223,7 @@
47223
47223
  "name": "maxNameReads",
47224
47224
  "type": "number",
47225
47225
  "optional": true,
47226
- "default": 20
47226
+ "default": 50
47227
47227
  },
47228
47228
  {
47229
47229
  "name": "excludeStatusIds",
@@ -47261,7 +47261,7 @@
47261
47261
  },
47262
47262
  "metadata": {
47263
47263
  "purpose": "Get everything the node knows about one company in a single call: the company record, its locations, contacts and notes, its open tickets, and its projects — every part separately reported with caps.",
47264
- "usage": "args[0] is the company id (number); args[1] (optional) is the options object { contactLimit?: number (default 50), noteLimit?: number (default 50), ticketLimit?: number (default 50), projectLimit?: number (default 50), names?: boolean (default false — ON adds the opt-in name/label siblings — the company row (owner / created-by resource names), the openTickets rows (the same closed-scope ticket-field siblings as tickets.fullContext, incl. the ticket picklist labels), the projects rows (lead / creator resource names + the contract name — project picklists stay OUT of the closed label scope) — + the top-level `names` section; the siblings ride the existing default-on caches (NO new cache)), maxNameReads?: number (default 20 — the budget of COLD wire name reads: cache hits are free of the budget, lowerable never raised; over-budget ids serve name: null and the names section sets `truncated: true`, never an error) }. Wire: exactly 6 reads (companies.get + 5 companyID-filtered queries). With names: true, N added name reads (N = the distinct uncached ids, capped by maxNameReads) + at most ONE fieldCatalog labels read (zero wire on a warm fieldCache). The openTickets part: open = status NOT in the resolved closed set — the closed ids come from the tenant's Tickets status picklist via `tickets.statusSemantics` (one cache-aware field-metadata read, zero wire on a warm cache), falling back to id 5 with `source: 'assumed'` and a part `note` that discloses it; the part carries `closedStatus { closedIds, source, note }` and the always-filtered query applies the resolved leaf (a single `not-eq` for one closed id, `notIn` for several); an empty scoped query shows no records returned to this API user, not proven tenant-wide absence. A conservative visibility warning is attached only when its filter-aware predicate covers the call; non-empty rows are returned as data. A missing company returns NOT_FOUND.",
47264
+ "usage": "args[0] is the company id (number); args[1] (optional) is the options object { contactLimit?: number (default 50), noteLimit?: number (default 50), ticketLimit?: number (default 50), projectLimit?: number (default 50), names?: boolean (default false — ON adds the opt-in name/label siblings — the company row (owner / created-by resource names), the openTickets rows (the same closed-scope ticket-field siblings as tickets.fullContext, incl. the ticket picklist labels), the projects rows (lead / creator resource names + the contract name — project picklists stay OUT of the closed label scope) — + the top-level `names` section; the siblings ride the existing default-on caches (NO new cache)), maxNameReads?: number (default 50 — the budget of COLD wire name reads: cache hits are free of the budget, lowerable never raised; over-budget ids serve name: null and the names section sets `truncated: true`, never an error) }. Wire: exactly 6 reads (companies.get + 5 companyID-filtered queries). With names: true, up to N added name reads (N = the distinct uncached ids, capped by maxNameReads; two or more uncached ids of one record table share one batched id-in query) + at most ONE fieldCatalog labels read (zero wire on a warm fieldCache). The openTickets part: open = status NOT in the resolved closed set — the closed ids come from the tenant's Tickets status picklist via `tickets.statusSemantics` (one cache-aware field-metadata read, zero wire on a warm cache), falling back to id 5 with `source: 'assumed'` and a part `note` that discloses it; the part carries `closedStatus { closedIds, source, note }` and the always-filtered query applies the resolved leaf (a single `not-eq` for one closed id, `notIn` for several); an empty scoped query shows no records returned to this API user, not proven tenant-wide absence. A conservative visibility warning is attached only when its filter-aware predicate covers the call; non-empty rows are returned as data. A missing company returns NOT_FOUND.",
47265
47265
  "preferredWhen": "The single read for 'what do you know about this company' — replaces 6 raw reads + describes with one call. companies.get alone is just the record.",
47266
47266
  "related": [
47267
47267
  "companies.get",
@@ -47271,7 +47271,7 @@
47271
47271
  "tickets.query",
47272
47272
  "projects.query"
47273
47273
  ],
47274
- "args": "args[0] is the company id (number); args[1] (optional) is the options object { contactLimit?: number, noteLimit?: number, ticketLimit?: number, projectLimit?: number } (all default 50), names?: boolean (default false — the opt-in name/label siblings + the `names` section), maxNameReads?: number (default 20, lowerable never raised — the budget of cold wire name reads). Bounded reads only, with no mutations. Empty lists reflect the current API-user query result; verify a known id before inferring tenant-wide absence.",
47274
+ "args": "args[0] is the company id (number); args[1] (optional) is the options object { contactLimit?: number, noteLimit?: number, ticketLimit?: number, projectLimit?: number } (all default 50), names?: boolean (default false — the opt-in name/label siblings + the `names` section), maxNameReads?: number (default 50, lowerable never raised — the budget of cold wire name reads). Bounded reads only, with no mutations. Empty lists reflect the current API-user query result; verify a known id before inferring tenant-wide absence.",
47275
47275
  "untrustedFields": [
47276
47276
  "UDF1",
47277
47277
  "UDF10",
@@ -47792,7 +47792,7 @@
47792
47792
  "params": {
47793
47793
  "type": "object",
47794
47794
  "required": false,
47795
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
47795
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
47796
47796
  "properties": [
47797
47797
  {
47798
47798
  "name": "filter",
@@ -49028,7 +49028,7 @@
49028
49028
  "params": {
49029
49029
  "type": "object",
49030
49030
  "required": false,
49031
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
49031
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
49032
49032
  "properties": [
49033
49033
  {
49034
49034
  "name": "filter",
@@ -49622,7 +49622,7 @@
49622
49622
  "params": {
49623
49623
  "type": "object",
49624
49624
  "required": false,
49625
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
49625
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
49626
49626
  "properties": [
49627
49627
  {
49628
49628
  "name": "filter",
@@ -50126,7 +50126,7 @@
50126
50126
  "params": {
50127
50127
  "type": "object",
50128
50128
  "required": false,
50129
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
50129
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
50130
50130
  "properties": [
50131
50131
  {
50132
50132
  "name": "filter",
@@ -50792,7 +50792,7 @@
50792
50792
  "params": {
50793
50793
  "type": "object",
50794
50794
  "required": false,
50795
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
50795
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
50796
50796
  "properties": [
50797
50797
  {
50798
50798
  "name": "filter",
@@ -51241,7 +51241,7 @@
51241
51241
  "params": {
51242
51242
  "type": "object",
51243
51243
  "required": false,
51244
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
51244
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
51245
51245
  "properties": [
51246
51246
  {
51247
51247
  "name": "filter",
@@ -51925,7 +51925,7 @@
51925
51925
  "params": {
51926
51926
  "type": "object",
51927
51927
  "required": false,
51928
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
51928
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
51929
51929
  "properties": [
51930
51930
  {
51931
51931
  "name": "filter",
@@ -52612,7 +52612,7 @@
52612
52612
  "params": {
52613
52613
  "type": "object",
52614
52614
  "required": false,
52615
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
52615
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
52616
52616
  "properties": [
52617
52617
  {
52618
52618
  "name": "filter",
@@ -53287,7 +53287,7 @@
53287
53287
  "params": {
53288
53288
  "type": "object",
53289
53289
  "required": false,
53290
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
53290
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
53291
53291
  "properties": [
53292
53292
  {
53293
53293
  "name": "filter",
@@ -53763,7 +53763,7 @@
53763
53763
  "params": {
53764
53764
  "type": "object",
53765
53765
  "required": false,
53766
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
53766
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
53767
53767
  "properties": [
53768
53768
  {
53769
53769
  "name": "filter",
@@ -54331,7 +54331,7 @@
54331
54331
  "params": {
54332
54332
  "type": "object",
54333
54333
  "required": false,
54334
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
54334
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
54335
54335
  "properties": [
54336
54336
  {
54337
54337
  "name": "filter",
@@ -54876,7 +54876,7 @@
54876
54876
  "params": {
54877
54877
  "type": "object",
54878
54878
  "required": false,
54879
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
54879
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
54880
54880
  "properties": [
54881
54881
  {
54882
54882
  "name": "filter",
@@ -55325,7 +55325,7 @@
55325
55325
  "params": {
55326
55326
  "type": "object",
55327
55327
  "required": false,
55328
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
55328
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
55329
55329
  "properties": [
55330
55330
  {
55331
55331
  "name": "filter",
@@ -55934,7 +55934,7 @@
55934
55934
  "params": {
55935
55935
  "type": "object",
55936
55936
  "required": false,
55937
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
55937
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
55938
55938
  "properties": [
55939
55939
  {
55940
55940
  "name": "filter",
@@ -56436,7 +56436,7 @@
56436
56436
  "params": {
56437
56437
  "type": "object",
56438
56438
  "required": false,
56439
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
56439
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
56440
56440
  "properties": [
56441
56441
  {
56442
56442
  "name": "filter",
@@ -57131,7 +57131,7 @@
57131
57131
  "params": {
57132
57132
  "type": "object",
57133
57133
  "required": false,
57134
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
57134
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
57135
57135
  "properties": [
57136
57136
  {
57137
57137
  "name": "filter",
@@ -58132,7 +58132,7 @@
58132
58132
  "opts": {
58133
58133
  "type": "object",
58134
58134
  "required": false,
58135
- "description": "ConfigurationItemProfileOptions — { noteLimit?: number (default 50), relatedLimit?: number (default 50), subscriptionLimit?: number (default 50 — the subscriptions part cap; a full page sets capped: true), names?: boolean (default false — opt-in name/label siblings + the top-level `names` section), maxNameReads?: number (default 20 — cold wire name-read budget, lowerable never raised) }.",
58135
+ "description": "ConfigurationItemProfileOptions — { noteLimit?: number (default 50), relatedLimit?: number (default 50), subscriptionLimit?: number (default 50 — the subscriptions part cap; a full page sets capped: true), names?: boolean (default false — opt-in name/label siblings + the top-level `names` section), maxNameReads?: number (default 50 — cold wire name-read budget, lowerable never raised) }.",
58136
58136
  "properties": [
58137
58137
  {
58138
58138
  "name": "noteLimit",
@@ -58162,7 +58162,7 @@
58162
58162
  "name": "maxNameReads",
58163
58163
  "type": "number",
58164
58164
  "optional": true,
58165
- "default": 20
58165
+ "default": 50
58166
58166
  }
58167
58167
  ]
58168
58168
  }
@@ -58189,7 +58189,7 @@
58189
58189
  },
58190
58190
  "metadata": {
58191
58191
  "purpose": "Get the full profile of one configuration item in a single call: the item, its type, category and product records, its notes, related items, billing-product associations and subscriptions (software/licence terms keyed by configurationItemID).",
58192
- "usage": "args[0] is the configuration item id (number); args[1] (optional) is the options object { noteLimit?: number (default 50), relatedLimit?: number (default 50), subscriptionLimit?: number (default 50), names?: boolean (default false — ON adds the opt-in name siblings on the ci row (closed scope: contractIDName, companyIDName, productIDName and vendorIDName, where vendorID names a company — the CI row has no issue-listed picklist fields, so no labels) + the top-level `names` section; the siblings ride the existing default-on caches (NO new cache)), maxNameReads?: number (default 20 — the budget of COLD wire name reads, lowerable never raised; over-budget ids serve name: null and the names section sets `truncated: true`, never an error) }. Wire: <=8 reads (get + type + category + product + 4 item-scoped queries; a missing or dangling type/category/product FK skips its read and serves null with parts.<part> 'absent'; productID 0 is never read). The product and subscriptions parts degrade on a typed vendor fault (e.g. an API user without read access to Products or Subscriptions): product null with parts.product 'unavailable', subscriptions { items: [], capped: false } with parts.subscriptions 'unavailable', and the fault code under unavailable.<part> (present only when a part degraded); a fault on any other part rejects the profile. subscriptions is the Subscriptions query filtered configurationItemID eq the item id, capped at subscriptionLimit (capped: true on a full page — more may exist). With names: true, N added name reads (N = the distinct uncached FK ids, capped by maxNameReads; productIDName comes from the product part's outcome, with no second read). productAssociations contains the scoped query rows and capped flag, with blind: false on a successful read; a product create requirement does not imply that associations are unreadable. A missing configuration item returns NOT_FOUND.",
58192
+ "usage": "args[0] is the configuration item id (number); args[1] (optional) is the options object { noteLimit?: number (default 50), relatedLimit?: number (default 50), subscriptionLimit?: number (default 50), names?: boolean (default false — ON adds the opt-in name siblings on the ci row (closed scope: contractIDName, companyIDName, productIDName and vendorIDName, where vendorID names a company — the CI row has no issue-listed picklist fields, so no labels) + the top-level `names` section; the siblings ride the existing default-on caches (NO new cache)), maxNameReads?: number (default 50 — the budget of COLD wire name reads, lowerable never raised; over-budget ids serve name: null and the names section sets `truncated: true`, never an error) }. Wire: <=8 reads (get + type + category + product + 4 item-scoped queries; a missing or dangling type/category/product FK skips its read and serves null with parts.<part> 'absent'; productID 0 is never read). The product and subscriptions parts degrade on a typed vendor fault (e.g. an API user without read access to Products or Subscriptions): product null with parts.product 'unavailable', subscriptions { items: [], capped: false } with parts.subscriptions 'unavailable', and the fault code under unavailable.<part> (present only when a part degraded); a fault on any other part rejects the profile. subscriptions is the Subscriptions query filtered configurationItemID eq the item id, capped at subscriptionLimit (capped: true on a full page — more may exist). With names: true, up to N added name reads (N = the distinct uncached FK ids, capped by maxNameReads; two or more uncached ids of one table share one batched id-in query; productIDName comes from the product part's outcome, with no second read). productAssociations contains the scoped query rows and capped flag, with blind: false on a successful read; a product create requirement does not imply that associations are unreadable. A missing configuration item returns NOT_FOUND.",
58193
58193
  "preferredWhen": "The single read for 'what is this configuration item' — replaces 7-8 raw reads + describes with one call, including the item's subscription terms (start/end dates, vendor, period price/cost).",
58194
58194
  "related": [
58195
58195
  "configuration_items.get",
@@ -58202,7 +58202,7 @@
58202
58202
  "products.get",
58203
58203
  "subscriptions.query"
58204
58204
  ],
58205
- "args": "args[0] is the configuration item id (number); args[1] (optional) is the options object { noteLimit?: number (default 50), relatedLimit?: number (default 50), subscriptionLimit?: number (default 50), names?: boolean (default false — the opt-in contract/company/product/vendor name siblings on the ci row + the `names` section), maxNameReads?: number (default 20, lowerable never raised — the budget of cold wire name reads) }. <=8 bounded reads inside, zero mutating; product associations and subscriptions are returned from scoped vendor queries without a static inventory or visibility assumption; the product and subscriptions parts degrade to 'unavailable' (with the fault code) on a typed vendor fault.",
58205
+ "args": "args[0] is the configuration item id (number); args[1] (optional) is the options object { noteLimit?: number (default 50), relatedLimit?: number (default 50), subscriptionLimit?: number (default 50), names?: boolean (default false — the opt-in contract/company/product/vendor name siblings on the ci row + the `names` section), maxNameReads?: number (default 50, lowerable never raised — the budget of cold wire name reads) }. <=8 bounded reads inside, zero mutating; product associations and subscriptions are returned from scoped vendor queries without a static inventory or visibility assumption; the product and subscriptions parts degrade to 'unavailable' (with the fault code) on a typed vendor fault.",
58206
58206
  "untrustedFields": [
58207
58207
  "UDF1",
58208
58208
  "UDF10",
@@ -58259,7 +58259,7 @@
58259
58259
  "params": {
58260
58260
  "type": "object",
58261
58261
  "required": false,
58262
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
58262
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
58263
58263
  "properties": [
58264
58264
  {
58265
58265
  "name": "filter",
@@ -59039,7 +59039,7 @@
59039
59039
  "params": {
59040
59040
  "type": "object",
59041
59041
  "required": false,
59042
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
59042
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
59043
59043
  "properties": [
59044
59044
  {
59045
59045
  "name": "filter",
@@ -59692,7 +59692,7 @@
59692
59692
  "opts": {
59693
59693
  "type": "object",
59694
59694
  "required": false,
59695
- "description": "CompanyFullProfileOptions — { contactLimit?: number (default 50), noteLimit?: number (default 50), ticketLimit?: number (default 50), projectLimit?: number (default 50), locationLimit?: number (default 50, 1-500), fields?: PartFields (per-part field projection: keys locations | contacts | notes | projects | openTickets, each a non-empty list of exact vendor field names — a listed part serves ONLY those fields (IncludeFields on the wire), an absent part keeps the full model; an unknown part key or field name is refused pre-request, zero wire; with names: true the name siblings appear only for the id fields kept), names?: boolean (default false — opt-in name/label siblings + the top-level `names` section), maxNameReads?: number (default 20 — cold wire name-read budget, lowerable never raised), excludeStatusIds?: number[] (explicit closed ticket-status ids for openTickets, overriding the resolved picklist), allowAssumed?: boolean (default true; false throws STATUS_SEMANTICS_UNRESOLVED instead of using the assumed closed id 5) }.",
59695
+ "description": "CompanyFullProfileOptions — { contactLimit?: number (default 50), noteLimit?: number (default 50), ticketLimit?: number (default 50), projectLimit?: number (default 50), locationLimit?: number (default 50, 1-500), fields?: PartFields (per-part field projection: keys locations | contacts | notes | projects | openTickets, each a non-empty list of exact vendor field names — a listed part serves ONLY those fields (IncludeFields on the wire), an absent part keeps the full model; an unknown part key or field name is refused pre-request, zero wire; with names: true the name siblings appear only for the id fields kept), names?: boolean (default false — opt-in name/label siblings + the top-level `names` section), maxNameReads?: number (default 50 — cold wire name-read budget, lowerable never raised), excludeStatusIds?: number[] (explicit closed ticket-status ids for openTickets, overriding the resolved picklist), allowAssumed?: boolean (default true; false throws STATUS_SEMANTICS_UNRESOLVED instead of using the assumed closed id 5) }.",
59696
59696
  "properties": [
59697
59697
  {
59698
59698
  "name": "contactLimit",
@@ -59739,7 +59739,7 @@
59739
59739
  "name": "maxNameReads",
59740
59740
  "type": "number",
59741
59741
  "optional": true,
59742
- "default": 20
59742
+ "default": 50
59743
59743
  },
59744
59744
  {
59745
59745
  "name": "excludeStatusIds",
@@ -60315,7 +60315,7 @@
60315
60315
  "params": {
60316
60316
  "type": "object",
60317
60317
  "required": false,
60318
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
60318
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
60319
60319
  "properties": [
60320
60320
  {
60321
60321
  "name": "filter",
@@ -61110,7 +61110,7 @@
61110
61110
  "params": {
61111
61111
  "type": "object",
61112
61112
  "required": false,
61113
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
61113
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
61114
61114
  "properties": [
61115
61115
  {
61116
61116
  "name": "filter",
@@ -61665,7 +61665,7 @@
61665
61665
  "params": {
61666
61666
  "type": "object",
61667
61667
  "required": false,
61668
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
61668
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
61669
61669
  "properties": [
61670
61670
  {
61671
61671
  "name": "filter",
@@ -62211,7 +62211,7 @@
62211
62211
  "params": {
62212
62212
  "type": "object",
62213
62213
  "required": false,
62214
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
62214
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
62215
62215
  "properties": [
62216
62216
  {
62217
62217
  "name": "filter",
@@ -62843,7 +62843,7 @@
62843
62843
  "params": {
62844
62844
  "type": "object",
62845
62845
  "required": false,
62846
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
62846
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
62847
62847
  "properties": [
62848
62848
  {
62849
62849
  "name": "filter",
@@ -63525,7 +63525,7 @@
63525
63525
  "params": {
63526
63526
  "type": "object",
63527
63527
  "required": false,
63528
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
63528
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
63529
63529
  "properties": [
63530
63530
  {
63531
63531
  "name": "filter",
@@ -64062,7 +64062,7 @@
64062
64062
  "params": {
64063
64063
  "type": "object",
64064
64064
  "required": false,
64065
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
64065
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
64066
64066
  "properties": [
64067
64067
  {
64068
64068
  "name": "filter",
@@ -64572,7 +64572,7 @@
64572
64572
  "params": {
64573
64573
  "type": "object",
64574
64574
  "required": false,
64575
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
64575
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
64576
64576
  "properties": [
64577
64577
  {
64578
64578
  "name": "filter",
@@ -65181,7 +65181,7 @@
65181
65181
  "params": {
65182
65182
  "type": "object",
65183
65183
  "required": false,
65184
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
65184
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
65185
65185
  "properties": [
65186
65186
  {
65187
65187
  "name": "filter",
@@ -65630,7 +65630,7 @@
65630
65630
  "params": {
65631
65631
  "type": "object",
65632
65632
  "required": false,
65633
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
65633
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
65634
65634
  "properties": [
65635
65635
  {
65636
65636
  "name": "filter",
@@ -66188,7 +66188,7 @@
66188
66188
  "params": {
66189
66189
  "type": "object",
66190
66190
  "required": false,
66191
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
66191
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
66192
66192
  "properties": [
66193
66193
  {
66194
66194
  "name": "filter",
@@ -66734,7 +66734,7 @@
66734
66734
  "params": {
66735
66735
  "type": "object",
66736
66736
  "required": false,
66737
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
66737
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
66738
66738
  "properties": [
66739
66739
  {
66740
66740
  "name": "filter",
@@ -67280,7 +67280,7 @@
67280
67280
  "params": {
67281
67281
  "type": "object",
67282
67282
  "required": false,
67283
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
67283
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
67284
67284
  "properties": [
67285
67285
  {
67286
67286
  "name": "filter",
@@ -68266,7 +68266,7 @@
68266
68266
  "params": {
68267
68267
  "type": "object",
68268
68268
  "required": false,
68269
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
68269
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
68270
68270
  "properties": [
68271
68271
  {
68272
68272
  "name": "filter",
@@ -68776,7 +68776,7 @@
68776
68776
  "params": {
68777
68777
  "type": "object",
68778
68778
  "required": false,
68779
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
68779
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
68780
68780
  "properties": [
68781
68781
  {
68782
68782
  "name": "filter",
@@ -69276,7 +69276,7 @@
69276
69276
  "params": {
69277
69277
  "type": "object",
69278
69278
  "required": false,
69279
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
69279
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
69280
69280
  "properties": [
69281
69281
  {
69282
69282
  "name": "filter",
@@ -69786,7 +69786,7 @@
69786
69786
  "params": {
69787
69787
  "type": "object",
69788
69788
  "required": false,
69789
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
69789
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
69790
69790
  "properties": [
69791
69791
  {
69792
69792
  "name": "filter",
@@ -70344,7 +70344,7 @@
70344
70344
  "params": {
70345
70345
  "type": "object",
70346
70346
  "required": false,
70347
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
70347
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
70348
70348
  "properties": [
70349
70349
  {
70350
70350
  "name": "filter",
@@ -70820,7 +70820,7 @@
70820
70820
  "opts": {
70821
70821
  "type": "object",
70822
70822
  "required": false,
70823
- "description": "CompanyFullProfileOptions — { contactLimit?: number (default 50), noteLimit?: number (default 50), ticketLimit?: number (default 50), projectLimit?: number (default 50), locationLimit?: number (default 50, 1-500), fields?: PartFields (per-part field projection: keys locations | contacts | notes | projects | openTickets, each a non-empty list of exact vendor field names — a listed part serves ONLY those fields (IncludeFields on the wire), an absent part keeps the full model; an unknown part key or field name is refused pre-request, zero wire; with names: true the name siblings appear only for the id fields kept), names?: boolean (default false — opt-in name/label siblings + the top-level `names` section), maxNameReads?: number (default 20 — cold wire name-read budget, lowerable never raised), excludeStatusIds?: number[] (explicit closed ticket-status ids for openTickets, overriding the resolved picklist), allowAssumed?: boolean (default true; false throws STATUS_SEMANTICS_UNRESOLVED instead of using the assumed closed id 5) }.",
70823
+ "description": "CompanyFullProfileOptions — { contactLimit?: number (default 50), noteLimit?: number (default 50), ticketLimit?: number (default 50), projectLimit?: number (default 50), locationLimit?: number (default 50, 1-500), fields?: PartFields (per-part field projection: keys locations | contacts | notes | projects | openTickets, each a non-empty list of exact vendor field names — a listed part serves ONLY those fields (IncludeFields on the wire), an absent part keeps the full model; an unknown part key or field name is refused pre-request, zero wire; with names: true the name siblings appear only for the id fields kept), names?: boolean (default false — opt-in name/label siblings + the top-level `names` section), maxNameReads?: number (default 50 — cold wire name-read budget, lowerable never raised), excludeStatusIds?: number[] (explicit closed ticket-status ids for openTickets, overriding the resolved picklist), allowAssumed?: boolean (default true; false throws STATUS_SEMANTICS_UNRESOLVED instead of using the assumed closed id 5) }.",
70824
70824
  "properties": [
70825
70825
  {
70826
70826
  "name": "contactLimit",
@@ -70867,7 +70867,7 @@
70867
70867
  "name": "maxNameReads",
70868
70868
  "type": "number",
70869
70869
  "optional": true,
70870
- "default": 20
70870
+ "default": 50
70871
70871
  },
70872
70872
  {
70873
70873
  "name": "excludeStatusIds",
@@ -71536,7 +71536,7 @@
71536
71536
  "params": {
71537
71537
  "type": "object",
71538
71538
  "required": false,
71539
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
71539
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
71540
71540
  "properties": [
71541
71541
  {
71542
71542
  "name": "filter",
@@ -72359,7 +72359,7 @@
72359
72359
  "params": {
72360
72360
  "type": "object",
72361
72361
  "required": false,
72362
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
72362
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
72363
72363
  "properties": [
72364
72364
  {
72365
72365
  "name": "filter",
@@ -73095,7 +73095,7 @@
73095
73095
  "params": {
73096
73096
  "type": "object",
73097
73097
  "required": false,
73098
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
73098
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
73099
73099
  "properties": [
73100
73100
  {
73101
73101
  "name": "filter",
@@ -73861,7 +73861,7 @@
73861
73861
  "params": {
73862
73862
  "type": "object",
73863
73863
  "required": false,
73864
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
73864
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
73865
73865
  "properties": [
73866
73866
  {
73867
73867
  "name": "filter",
@@ -74377,7 +74377,7 @@
74377
74377
  "params": {
74378
74378
  "type": "object",
74379
74379
  "required": false,
74380
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
74380
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
74381
74381
  "properties": [
74382
74382
  {
74383
74383
  "name": "filter",
@@ -74986,7 +74986,7 @@
74986
74986
  "params": {
74987
74987
  "type": "object",
74988
74988
  "required": false,
74989
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
74989
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
74990
74990
  "properties": [
74991
74991
  {
74992
74992
  "name": "filter",
@@ -75505,7 +75505,7 @@
75505
75505
  "params": {
75506
75506
  "type": "object",
75507
75507
  "required": false,
75508
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
75508
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
75509
75509
  "properties": [
75510
75510
  {
75511
75511
  "name": "filter",
@@ -76187,7 +76187,7 @@
76187
76187
  "params": {
76188
76188
  "type": "object",
76189
76189
  "required": false,
76190
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
76190
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
76191
76191
  "properties": [
76192
76192
  {
76193
76193
  "name": "filter",
@@ -76636,7 +76636,7 @@
76636
76636
  "params": {
76637
76637
  "type": "object",
76638
76638
  "required": false,
76639
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
76639
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
76640
76640
  "properties": [
76641
76641
  {
76642
76642
  "name": "filter",
@@ -77138,7 +77138,7 @@
77138
77138
  "opts": {
77139
77139
  "type": "object",
77140
77140
  "required": false,
77141
- "description": "CompositeContextOptions — { names?: boolean (default false — the opt-in name siblings on the project row: the creator resource name + the contract name — project picklists stay label-less — + the top-level `names` section; the siblings ride the existing default-on caches), maxNameReads?: number (default 20 — the budget of COLD wire name reads, lowerable never raised; over-budget ids serve name: null + capped: true, never an error) }",
77141
+ "description": "CompositeContextOptions — { names?: boolean (default false — the opt-in name siblings on the project row: the creator resource name + the contract name — project picklists stay label-less — + the top-level `names` section; the siblings ride the existing default-on caches), maxNameReads?: number (default 50 — the budget of COLD wire name reads, lowerable never raised; over-budget ids serve name: null + capped: true, never an error) }",
77142
77142
  "properties": [
77143
77143
  {
77144
77144
  "name": "names",
@@ -77150,7 +77150,7 @@
77150
77150
  "name": "maxNameReads",
77151
77151
  "type": "number",
77152
77152
  "optional": true,
77153
- "default": 20
77153
+ "default": 50
77154
77154
  }
77155
77155
  ]
77156
77156
  }
@@ -77176,12 +77176,12 @@
77176
77176
  },
77177
77177
  "metadata": {
77178
77178
  "purpose": "Get a project with its primary context records in one call.",
77179
- "usage": "Each part is a single-record fetch; no sub-lists, no limit. expand: true returns full records. args[1] (optional): { names?: boolean (default false — ON adds the opt-in name siblings on the project row (closed scope: the creator resource name + the contract name — project picklists stay label-less) + the top-level `names` section; the siblings ride the existing default-on caches (NO new cache)), maxNameReads?: number (default 20 — the budget of COLD wire name reads, lowerable never raised; over-budget ids serve name: null and the names section sets `truncated: true`, never an error) }.",
77179
+ "usage": "Each part is a single-record fetch; no sub-lists, no limit. expand: true returns full records. args[1] (optional): { names?: boolean (default false — ON adds the opt-in name siblings on the project row (closed scope: the creator resource name + the contract name — project picklists stay label-less) + the top-level `names` section; the siblings ride the existing default-on caches (NO new cache)), maxNameReads?: number (default 50 — the budget of COLD wire name reads, lowerable never raised; over-budget ids serve name: null and the names section sets `truncated: true`, never an error) }.",
77180
77180
  "preferredWhen": null,
77181
77181
  "related": [
77182
77182
  "projects.get"
77183
77183
  ],
77184
- "args": "args[1] (optional): { names?: boolean (default false — the opt-in name siblings on the project row + the `names` section), maxNameReads?: number (default 20, lowerable never raised) }. No-arg calls are unchanged (1..4 wire).",
77184
+ "args": "args[1] (optional): { names?: boolean (default false — the opt-in name siblings on the project row + the `names` section), maxNameReads?: number (default 50, lowerable never raised) }. No-arg calls are unchanged (1..4 wire).",
77185
77185
  "untrustedFields": [
77186
77186
  "UDF1",
77187
77187
  "UDF10",
@@ -77455,7 +77455,7 @@
77455
77455
  "params": {
77456
77456
  "type": "object",
77457
77457
  "required": false,
77458
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
77458
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
77459
77459
  "properties": [
77460
77460
  {
77461
77461
  "name": "filter",
@@ -78241,7 +78241,7 @@
78241
78241
  "params": {
78242
78242
  "type": "object",
78243
78243
  "required": false,
78244
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
78244
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
78245
78245
  "properties": [
78246
78246
  {
78247
78247
  "name": "filter",
@@ -78858,7 +78858,7 @@
78858
78858
  "params": {
78859
78859
  "type": "object",
78860
78860
  "required": false,
78861
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
78861
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
78862
78862
  "properties": [
78863
78863
  {
78864
78864
  "name": "filter",
@@ -79426,7 +79426,7 @@
79426
79426
  "params": {
79427
79427
  "type": "object",
79428
79428
  "required": false,
79429
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
79429
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
79430
79430
  "properties": [
79431
79431
  {
79432
79432
  "name": "filter",
@@ -80332,7 +80332,7 @@
80332
80332
  "params": {
80333
80333
  "type": "object",
80334
80334
  "required": false,
80335
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
80335
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
80336
80336
  "properties": [
80337
80337
  {
80338
80338
  "name": "filter",
@@ -80985,7 +80985,7 @@
80985
80985
  "params": {
80986
80986
  "type": "object",
80987
80987
  "required": false,
80988
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
80988
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
80989
80989
  "properties": [
80990
80990
  {
80991
80991
  "name": "filter",
@@ -81461,7 +81461,7 @@
81461
81461
  "params": {
81462
81462
  "type": "object",
81463
81463
  "required": false,
81464
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
81464
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
81465
81465
  "properties": [
81466
81466
  {
81467
81467
  "name": "filter",
@@ -81908,7 +81908,7 @@
81908
81908
  "params": {
81909
81909
  "type": "object",
81910
81910
  "required": false,
81911
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
81911
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
81912
81912
  "properties": [
81913
81913
  {
81914
81914
  "name": "filter",
@@ -82443,7 +82443,7 @@
82443
82443
  "params": {
82444
82444
  "type": "object",
82445
82445
  "required": false,
82446
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
82446
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
82447
82447
  "properties": [
82448
82448
  {
82449
82449
  "name": "filter",
@@ -82972,7 +82972,7 @@
82972
82972
  "params": {
82973
82973
  "type": "object",
82974
82974
  "required": false,
82975
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
82975
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
82976
82976
  "properties": [
82977
82977
  {
82978
82978
  "name": "filter",
@@ -83603,7 +83603,7 @@
83603
83603
  "params": {
83604
83604
  "type": "object",
83605
83605
  "required": false,
83606
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
83606
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
83607
83607
  "properties": [
83608
83608
  {
83609
83609
  "name": "filter",
@@ -84361,7 +84361,7 @@
84361
84361
  "params": {
84362
84362
  "type": "object",
84363
84363
  "required": false,
84364
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
84364
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
84365
84365
  "properties": [
84366
84366
  {
84367
84367
  "name": "filter",
@@ -84898,7 +84898,7 @@
84898
84898
  "params": {
84899
84899
  "type": "object",
84900
84900
  "required": false,
84901
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
84901
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
84902
84902
  "properties": [
84903
84903
  {
84904
84904
  "name": "filter",
@@ -85512,7 +85512,7 @@
85512
85512
  "params": {
85513
85513
  "type": "object",
85514
85514
  "required": false,
85515
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
85515
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
85516
85516
  "properties": [
85517
85517
  {
85518
85518
  "name": "filter",
@@ -86076,7 +86076,7 @@
86076
86076
  "params": {
86077
86077
  "type": "object",
86078
86078
  "required": false,
86079
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
86079
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
86080
86080
  "properties": [
86081
86081
  {
86082
86082
  "name": "filter",
@@ -86723,7 +86723,7 @@
86723
86723
  "params": {
86724
86724
  "type": "object",
86725
86725
  "required": false,
86726
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
86726
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
86727
86727
  "properties": [
86728
86728
  {
86729
86729
  "name": "filter",
@@ -87436,7 +87436,7 @@
87436
87436
  "params": {
87437
87437
  "type": "object",
87438
87438
  "required": false,
87439
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
87439
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
87440
87440
  "properties": [
87441
87441
  {
87442
87442
  "name": "filter",
@@ -88324,7 +88324,7 @@
88324
88324
  "params": {
88325
88325
  "type": "object",
88326
88326
  "required": false,
88327
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
88327
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
88328
88328
  "properties": [
88329
88329
  {
88330
88330
  "name": "filter",
@@ -88933,7 +88933,7 @@
88933
88933
  "params": {
88934
88934
  "type": "object",
88935
88935
  "required": false,
88936
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
88936
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
88937
88937
  "properties": [
88938
88938
  {
88939
88939
  "name": "filter",
@@ -89491,7 +89491,7 @@
89491
89491
  "params": {
89492
89492
  "type": "object",
89493
89493
  "required": false,
89494
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
89494
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
89495
89495
  "properties": [
89496
89496
  {
89497
89497
  "name": "filter",
@@ -90305,7 +90305,7 @@
90305
90305
  "opts": {
90306
90306
  "type": "object",
90307
90307
  "required": false,
90308
- "description": "TicketFullContextOptions — { noteLimit?: number (default 50), timeEntryLimit?: number (default 50), includeAttachments?: boolean (default true — metadata only), includeHistories?: boolean (default false), textLimit?: number (default 500 — long text truncated in the served copy, flagged), names?: boolean (default false — opt-in name/label siblings + the top-level `names` section), maxNameReads?: number (default 20 — cold wire name-read budget, lowerable never raised) }.",
90308
+ "description": "TicketFullContextOptions — { noteLimit?: number (default 50), timeEntryLimit?: number (default 50), includeAttachments?: boolean (default true — metadata only), includeHistories?: boolean (default false), textLimit?: number (default 500 — long text truncated in the served copy, flagged), names?: boolean (default false — opt-in name/label siblings + the top-level `names` section), maxNameReads?: number (default 50 — cold wire name-read budget, lowerable never raised) }.",
90309
90309
  "properties": [
90310
90310
  {
90311
90311
  "name": "noteLimit",
@@ -90347,7 +90347,7 @@
90347
90347
  "name": "maxNameReads",
90348
90348
  "type": "number",
90349
90349
  "optional": true,
90350
- "default": 20
90350
+ "default": 50
90351
90351
  }
90352
90352
  ]
90353
90353
  }
@@ -90374,7 +90374,7 @@
90374
90374
  },
90375
90375
  "metadata": {
90376
90376
  "purpose": "Get the full context of one ticket in a single call: the ticket, its company, contact, assigned resource and project, plus its notes, time entries, attachment metadata and (opt-in) histories — every part separately reported.",
90377
- "usage": "args[0] is the ticket id (number); args[1] (optional) is the options object { noteLimit?: number (default 50, the per-part record cap), timeEntryLimit?: number (default 50), includeAttachments?: boolean (default true — metadata only: ids + names, data download stays a separate op), includeHistories?: boolean (default false — budget control), textLimit?: number (default 500 — note/history long text is truncated in the served copy and flagged), names?: boolean (default false — ON adds the opt-in name/label siblings on the ticket row and each time-entry row (closed scope: the ticket picklist labels status/priority/ticketType/issueType/subIssueType/source/queue + the contract/billingCode/role record-FK names + the creator + time-entry resource names) + the top-level `names` section, and — when the ticket (or, failing that, its contract) carries an `organizationalLevelAssociationID` — the Line of Business chain (association -> OrganizationalLevel2 (the customizable \"Line of Business\" label) -> OrganizationalLevel1 (the division)) as the top-level `lineOfBusiness` field: 1 cached contract read when the ticket has no pointer, then 1 association read + up to 2 level reads in parallel, inside the same maxNameReads budget (an absent pointer reports all-`missing` markers; a dangling record reports `unresolved`; a broken chain is never an error); the siblings ride the existing default-on caches — the read cache / field cache, NO new cache), maxNameReads?: number (default 20 — the budget of COLD wire name reads: cache hits are free of the budget, the caller may lower it, never raise it; over-budget ids serve name: null and the names section sets `truncated: true` — never an error) }. Wire: 8 bounded reads (the getContext 5 parts + notes + time entries + attachments), 10 with includeHistories, in a zone where the child routes work — a zone that drops a child route (404/500) recovers that part on the plan-declared scoped read (an extra wire per recovered part; the part status records the route: 'ok-scoped', or 'miss' with the route facts when every route failed). Per-part status in `parts` (a missing/dangling FK part is null + 'absent' — a part miss never fails the op; the ticket itself missing throws NOT_FOUND). PII fields ride the redaction choke point; the blind-surface sentinel rides the pinned zone facts. With names: true, N added name/label reads (N = the distinct uncached ids, capped by maxNameReads; the label channel is ONE fieldCatalog labels:true read — zero extra wire on a warm fieldCache).",
90377
+ "usage": "args[0] is the ticket id (number); args[1] (optional) is the options object { noteLimit?: number (default 50, the per-part record cap), timeEntryLimit?: number (default 50), includeAttachments?: boolean (default true — metadata only: ids + names, data download stays a separate op), includeHistories?: boolean (default false — budget control), textLimit?: number (default 500 — note/history long text is truncated in the served copy and flagged), names?: boolean (default false — ON adds the opt-in name/label siblings on the ticket row and each time-entry row (closed scope: the ticket picklist labels status/priority/ticketType/issueType/subIssueType/source/queue + the contract/billingCode/role record-FK names + the creator + time-entry resource names) + the top-level `names` section, and — when the ticket (or, failing that, its contract) carries an `organizationalLevelAssociationID` — the Line of Business chain (association -> OrganizationalLevel2 (the customizable \"Line of Business\" label) -> OrganizationalLevel1 (the division)) as the top-level `lineOfBusiness` field: 1 cached contract read when the ticket has no pointer, then 1 association read + up to 2 level reads in parallel, inside the same maxNameReads budget (an absent pointer reports all-`missing` markers; a dangling record reports `unresolved`; a broken chain is never an error); the siblings ride the existing default-on caches — the read cache / field cache, NO new cache), maxNameReads?: number (default 50 — the budget of COLD wire name reads: cache hits are free of the budget, the caller may lower it, never raise it; over-budget ids serve name: null and the names section sets `truncated: true` — never an error) }. Wire: 8 bounded reads (the getContext 5 parts + notes + time entries + attachments), 10 with includeHistories, in a zone where the child routes work — a zone that drops a child route (404/500) recovers that part on the plan-declared scoped read (an extra wire per recovered part; the part status records the route: 'ok-scoped', or 'miss' with the route facts when every route failed). Per-part status in `parts` (a missing/dangling FK part is null + 'absent' — a part miss never fails the op; the ticket itself missing throws NOT_FOUND). PII fields ride the redaction choke point; the blind-surface sentinel rides the pinned zone facts. With names: true, up to N added name/label reads (N = the distinct uncached ids, capped by maxNameReads; two or more uncached ids of one record table share one batched id-in query; the label channel is ONE fieldCatalog labels:true read — zero extra wire on a warm fieldCache).",
90378
90378
  "preferredWhen": "The single read for 'tell me everything about this ticket' — replaces 9 raw relationship reads + describes with one call. tickets.getContext is the lightweight 5-part subset; tickets.timeline is the merged chronological event view.",
90379
90379
  "related": [
90380
90380
  "tickets.getContext",
@@ -90383,7 +90383,7 @@
90383
90383
  "tickets.findByTicketNumber",
90384
90384
  "tickets.lineOfBusiness"
90385
90385
  ],
90386
- "args": "args[0] is the ticket id (number); args[1] (optional) is the options object { noteLimit?: number (default 50), timeEntryLimit?: number (default 50), includeAttachments?: boolean (default true), includeHistories?: boolean (default false), textLimit?: number (default 500), names?: boolean (default false — the opt-in name/label siblings + the `names` section + the top-level `lineOfBusiness` chain (1 association + up to 2 level reads, same budget; when the ticket has no pointer of its own the ticket -> contract fallback adds 1 cached contract read)), maxNameReads?: number (default 20 — the budget of COLD wire name reads, lowerable never raised) }. 8-10 bounded reads inside a working zone (a zone that drops a child route recovers it on the plan-declared scoped read — the per-part status records the route), zero mutating; per-part `parts` status; dangling FK → null part (never an error); ticket missing → NOT_FOUND.",
90386
+ "args": "args[0] is the ticket id (number); args[1] (optional) is the options object { noteLimit?: number (default 50), timeEntryLimit?: number (default 50), includeAttachments?: boolean (default true), includeHistories?: boolean (default false), textLimit?: number (default 500), names?: boolean (default false — the opt-in name/label siblings + the `names` section + the top-level `lineOfBusiness` chain (1 association + up to 2 level reads, same budget; when the ticket has no pointer of its own the ticket -> contract fallback adds 1 cached contract read)), maxNameReads?: number (default 50 — the budget of COLD wire name reads, lowerable never raised) }. 8-10 bounded reads inside a working zone (a zone that drops a child route recovers it on the plan-declared scoped read — the per-part status records the route), zero mutating; per-part `parts` status; dangling FK → null part (never an error); ticket missing → NOT_FOUND.",
90387
90387
  "untrustedFields": [
90388
90388
  "UDF1",
90389
90389
  "UDF10",
@@ -90954,7 +90954,7 @@
90954
90954
  "params": {
90955
90955
  "type": "object",
90956
90956
  "required": false,
90957
- "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + up to 20 cold name reads per call; the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the 20 uncached name-read cap or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
90957
+ "description": "The query params object (CLOSED — unknown keys are refused pre-request with a typed error naming the key + the valid keys): { filter?: FilterCondition[], limit?: number (total cap, default 500, max 2500 — NOT a page size), latest?: number (1-2500; the N highest-id rows — the vendor has no sort; count-probe + id-threshold semantics, see docs/API.md §4; the same-process latest-walk count probes are briefly cached per client — a 30 s TTL, cleared on a write through this client (a probe in flight across the write is discarded by the cache's invalidation epoch); the plain count op and aggregate stay fresh; limit and latest may coexist only when limit >= latest), noteChars?: number (integer >= 1; opt-in per-field char cap on the untrusted-class free-text fields of the resource (notes/descriptions/UDF prose): each capped value = the first N chars + the explicit marker … [+N chars truncated] (N = chars removed), applied at result-build BEFORE the server-side untrusted wrapper; absent = the raw, untruncated result), includeFields?: string[], fieldCheck?: boolean (default true: every includeFields NAME and every filter condition FIELD NAME is checked pre-request against the live field set (the vendored spec when that read faults — the resource's plan-pinned live-only fields, metadata.liveOnlyFields, are exempt from the vendored refusal: the live entity carries them and the vendor model omits them (e.g. time_entries postedDate/postedOnTime) — a truly unknown name is still refused); a refusal costs zero QUERY wire — the query POST and the count probe that anchors latest are never issued, a COLD fieldCache performs one authority metadata GET first (zero on a warm one); the one udf:true condition per query is exempt (a tenant UDF name is absent from the vendored spec by definition); false disables both name checks, never the shape/grammar checks; the count verb's bag is name-checked the same way and carries the same fieldCheck opt-out; a direct queryBounded/queryAll call is not checked (no bag contract there)), names?: boolean (absent = the client's labels option — OFF by default (the label pass costs a field-catalog read + a capped number of cold name reads per call (several ids of one table share one batched read); the MCP server (iit-mcp-autotask) enables the client labels option when it re-pins to this version (downstream)); true adds the row label siblings: picklist <field>Label (statusLabel), <fk>Name for companyID/contractID/ticketID/roleID/billingCodeID/productID/vendorID (a company)/configurationItemCategoryID/*ResourceID (contractIDName, productIDName, assignedResourceIDName) — null, never missing, when the field is keyed with a null value, a 0 id (except companyID/vendorID 0, which name the tenant owner's own company) or an unresolved value (a picklist 0 is labelled when the picklist defines it); a names summary appears only when something was left unlabelled (the uncached name-read cap, an id a batched name read did not return, or an unresolved picklist value); false = raw ids, ~30% smaller on a full tickets row), via?: \"post\" | \"get\" (vendor GET has a 2048-char cap), acknowledgeUnscoped?: boolean (query verb only; a walk expected to read over 500 rows with no id scope in the filter is guarded — one pre-flight <resource>.count, and a count over 500 refuses with the typed SCOPE_REQUIRED before the walk; true walks the unscoped set anyway; a count that fails or is blind lets the walk run and the result's scopeNote says so), walk?: object (walk.maxRows: positive integer — (D6): the walk's TOTAL row cap — serves at most maxRows rows in total; the page size stays min(500, limit), so one vendor page may fetch more rows than maxRows and the surplus is sliced off (fetched, not served); the result's truncated flag — false only when the cursor is exhausted, true iff more rows exist past the cap (so an exact-multiple count with a live next page reports true); not combinable with latest) }. Filter grammar: 13 operators (eq, noteq, gt, gte, lt, lte, beginsWith, endsWith, contains, exist, notExist, in, notIn; in/notIn: array or \"1,2,3\") + and/or grouping (no not-group); field names in vendor API casing (via operations.fieldCatalog); caps: 500 OR-conditions / 1 UDF / 8000 chars.",
90958
90958
  "properties": [
90959
90959
  {
90960
90960
  "name": "filter",
@@ -91788,8 +91788,8 @@
91788
91788
  ]
91789
91789
  },
91790
91790
  "operations": {
91791
- "planHash": "929593354ffa904a038cb7cbbf7b91ffbb05e9f06078181e9f225505f3463a88",
91792
- "builtAt": "2026-11-20T10:01:10.493Z",
91791
+ "planHash": "c97deac4ac7c3ca9ce4cea10ac6a50bb19b6517993d9657c52397964e1fcf715",
91792
+ "builtAt": "2026-11-20T11:02:13.904Z",
91793
91793
  "records": [
91794
91794
  {
91795
91795
  "id": "operations.aggregate",