jaz-clio 5.74.1 → 6.0.0

package/README.md

@@ -3,13 +3,13 @@
 <p align="center">
   <a href="https://www.npmjs.com/package/jaz-clio"><img src="https://img.shields.io/npm/v/jaz-clio?style=for-the-badge&logo=npm" alt="npm"></a>
   <a href="https://www.npmjs.com/package/jaz-clio"><img src="https://img.shields.io/npm/dm/jaz-clio?style=for-the-badge&label=downloads" alt="npm downloads"></a>
-  <img src="https://img.shields.io/badge/tools-386-blue?style=for-the-badge" alt="386 tools">
+  <img src="https://img.shields.io/badge/tools-381-blue?style=for-the-badge" alt="381 tools">
   <a href="https://github.com/teamtinvio/jaz-ai/blob/main/LICENSE"><img src="https://img.shields.io/github/license/teamtinvio/jaz-ai?style=for-the-badge&color=green" alt="License"></a>
 </p>
 
 <p align="center"><b>Jaz accounting on the command line, and inside any AI agent.</b></p>
 
-386 tools · 76 command groups · 7 skills · 13 calculators · 12 close playbooks · 160 field-tested API rules.
+381 tools · 74 command groups · 7 skills · 13 calculators · 12 close playbooks · 148 field-tested API rules.
 
 ```bash
 npm install -g jaz-clio
@@ -58,7 +58,7 @@ clio invoices create --contact "ACME" --json           # draft an invoice, JSON
 clio bank import --file statement.csv --account <id>    # import and auto-reconcile
 clio exports download --type profit-and-loss --format PDF  # download the P&L as a PDF
 clio calc lease --payment 5000 --term 36 --rate 5      # IFRS 16, offline, instant
-clio jobs month-end --period 2026-03                   # step-by-step close playbook
+clio jobs bank-recon match --input bank-data.json      # bank reconciliation matcher, offline
 clio magic create --file receipt.pdf                   # AI extracts, drafts the transaction
 clio invoices search --query 'status:unpaid AND $500+' # structured per-entity search
 clio ledger-find-fix preview --level TRANSACTIONS --input recode.json  # find & fix (recode) across record types
@@ -70,17 +70,17 @@ clio modules list                                      # which features are swit
 clio navigate reports.profit-and-loss                  # deep link into the Jaz dashboard (offline)
 ```
 
-76 command groups, 17 report types, 13 calculators, 12 job playbooks. Every command takes `--json`. Run `clio --help` for the full list.
+74 command groups, 17 report types, 13 calculators, 12 job playbooks. Every command takes `--json`. Run `clio --help` for the full list.
 
 Command groups by area:
 
 - **Sales and purchases**: `invoices`, `bills`, `customer-credit-notes`, `supplier-credit-notes`, `sale-orders`, `purchase-orders`, `payments`, `unapplied-payments`, `approvals`, `drafts`, `schedulers`, `subscriptions`
 - **Ledger and banking**: `journals`, `cash-in`, `cash-out`, `cash-transfer`, `cashflow`, `bank`, `bank-rules`, `reconciliations` (`recon`), `ledger-find-fix`, `quick-fix`, `fixed-assets` (`fa`)
-- **Master data**: `accounts`, `contacts`, `contact-groups`, `items`, `purchase-items`, `catalogs`, `inventory` (`inv`), `tags`, `custom-fields`, `nano-classifiers`, `tax-profiles`, `currencies`, `currency-rates`, `capsules`, `capsule-recipes`
+- **Master data**: `accounts`, `contacts`, `contact-groups`, `items`, `purchase-items`, `catalogs`, `inventory` (`inv`), `tags`, `custom-fields`, `nano-classifiers`, `tax-profiles`, `currencies`, `currency-rates`, `capsules`
 - **Claims**: `claims`, `claim-types`, `claim-profiles`, `posting-rules`, `employees`
 - **Reports and data**: `reports`, `exports`, `export-records`, `report-templates`, `pseudo-sql` (`sql`), `filing-submissions`, `background-jobs`
 - **Organization**: `org`, `org-users`, `references`, `modules` (`features`), `bookmarks`, `attachments`, `magic`, `jots`, `navigate` (`nav`), `help-center` (`hc`)
-- **Offline**: `calc`, `capsule-transaction` (`ct`), `jobs`
+- **Calculators and job tools**: `calc` (offline), `jobs`
 - **Setup and tooling**: `auth`, `init`, `update`, `versions`, `version`, `health`, `completion`, `context`, `schema`, `resolve`, `mcp`, `mcp-call`, `serve`
 
 ### Foreign currency
@@ -108,7 +108,7 @@ that ends up in a journal.
 
 ## MCP server
 
-386 tools for any AI agent that speaks MCP. Runs locally: no cloud, no ports.
+381 tools for any AI agent that speaks MCP. Runs locally: no cloud, no ports.
 
 > **No install at all?** Claude.ai, ChatGPT, Cowork, and Microsoft Copilot Studio can use Jaz through the hosted connector. Add `https://mcp.jaz.ai/mcp` as a custom connector and sign in with OAuth, no key. The local setup below is for terminal use, scripting, and editors that run MCP servers as local processes.
 
@@ -141,7 +141,7 @@ OAuth can reach the organizations granted at sign-in; `--org` pins one, and with
 
 ## Skills
 
-160 API rules from production testing: field-name maps, error-recovery patterns, response-shape quirks, plus 12 job playbooks. Installable into any agent project, no server involved.
+148 API rules from production testing: field-name maps, error-recovery patterns, response-shape quirks, plus 12 job playbooks. Installable into any agent project, no server involved.
 
 ```bash
 clio init                     # auto-detect the agent, install skills + agent-rules

package/assets/skills/api/SKILL.md

@@ -1,9 +1,9 @@
 ---
 name: jaz-api
-version: 5.74.1
+version: 6.0.0
 description: >-
   Use this skill whenever you call, debug, or review code that touches the Jaz
-  REST API. Covers field names, response shapes, 160 production gotchas, error
+  REST API. Covers field names, response shapes, 148 production gotchas, error
   recovery (422/400/404/500), search filters, pagination, and edge cases for
   every endpoint: invoices, bills, credit notes, journals, cash entries,
   payments, contacts, CoA, items, tax profiles, bank records, fixed assets,
@@ -171,7 +171,7 @@ The rest of this skill (field names, gotchas, error catalog, dependency order, s
 43. **Search sort is an object**: `{ sort: { sortBy: ["valueDate"], order: "DESC" } }`. Required when `offset` is present (even `offset: 0`).
 44. **Bank records**. **Create**: Multipart CSV/OFX via `POST /magic/importBankStatementFromAttachment` or JSON via `POST /bank-records/:accountResourceId` with `{ records: [{amount, transactionDate, description?, payerOrPayee?, reference?}] }` (positive = cash-in, negative = cash-out, response: `{data: {errors: []}}`). **Search**: `POST /bank-records/:accountResourceId/search`; filter fields: `valueDate` (DateExpression), `status` (StringExpression: UNRECONCILED, RECONCILED, ARCHIVED, POSSIBLE_DUPLICATE), `description`, `extContactName` (payer/payee), `extReference`, `netAmount` (BigDecimalExpression), `extAccountNumber`. Sort by `valueDate` DESC default.
 45. **Withholding tax** on bills/supplier CNs only. Retry pattern: if `WITHHOLDING_CODE_NOT_FOUND`, strip field and retry.
-46. **Known API bugs (500s)**: capsules POST (upstream returns nil). **Catalogs POST, inventory balances by status GET, contact groups PUT and custom fields PUT are NO LONGER on this list.** Catalogs POST returned 201 on a live probe 2026-09-23 (item `unit` omitted, it is optional); its response is counts (`createdCatalogs`, `createdCatalogItems`), not a resourceId, so find the new catalog with `search_catalogs`. Inventory balances by status works (Rule 97). Contact groups PUT and custom fields PUT were re-probed live 2026-09-02 and returned 200 and applied the change. Do not route around any of them.
+46. **Known API bugs (500s)**: none on this list today. **Capsules POST, catalogs POST, inventory balances by status GET, contact groups PUT and custom fields PUT are NO LONGER on this list.** Capsules POST creates the capsule and returns its `resourceId`. Catalogs POST returned 201 on a live probe 2026-09-23 (item `unit` omitted, it is optional); its response is counts (`createdCatalogs`, `createdCatalogItems`), not a resourceId, so find the new catalog with `search_catalogs`. Inventory balances by status works (Rule 97). Contact groups PUT and custom fields PUT were re-probed live 2026-09-02 and returned 200 and applied the change. Do not route around any of them.
 47. **Non-existent endpoints**: `POST /inventory/adjustments` (and `/inventory-adjustments`, `/inventory-items/:id/adjustments`, `/items/:id/inventory-adjustments`; no stock-adjustment write path exists at any spelling), `GET /payments` (list), and `POST /payments/search` return 404; these endpoints are not implemented. For payments, per-payment CRUD is `GET/PUT/DELETE /payments/:resourceId` and there ARE two payment-scoped searches; see Rule 64. `POST /deposits` also 404s, but for a different reason, see Rule 47a; do not read it as a missing feature.
 47a. **Deposits are not an entity; the 404 on `/deposits` is by design.** A deposit is a **flag on a Chart of Accounts account**, not a document: the account carries `depositContactType` = `CUSTOMER` (advance received, a liability) or `SUPPLIER` (advance paid, an asset). Once flagged, a deposit movement is an **ordinary transaction against that account**: top up with `POST /journals` / `POST /cash-in-entries` / `POST /cash-out-entries` (one leg on the flagged account), draw down with `POST /invoices/:id/payments` or `POST /bills/:id/payments` passing `accountResourceId` = the flagged account **and `paymentMethod` set to something other than BANK_TRANSFER / CASH / CHEQUE** (use `OTHER`; those three force a bank/cash account and 422 here, see Rule 80), read with `POST /cashflow-transactions/search`. **Flagging the account is a web-app action**: `depositContactType` is on no chart-of-accounts request or response model here, and the platform-backend mutation that sets it (`configureDepositAccounts`) is not proxied, so `POST /chart-of-accounts` cannot create a deposit account. Over the API you can only read and post against an account someone already flagged. Full walkthrough: `references/endpoints.md` → Deposits.
 48. **Attachments (full CRUD)**: **Add**: `POST /:type/:id/attachments` (multipart, `file` field, `application/pdf` or `image/*`, NOT `text/plain`). **List**: `GET /:type/:id/attachments`. **Delete**: `DELETE /:type/:id/attachments/:attachmentResourceId` (HTTP 200). CLI: `clio attachments add --file <path>` or `--url <url>`, `clio attachments list`, `clio attachments delete <attachmentResourceId>`. **The list response shape depends on the entity**. Measured raw 2026-09-07: `/invoices/:id/attachments` returns `{ reference, resourceId, attachments: [{fileName, fileType, fileId, attachmentResourceId}] }`, while `/bills/:id/attachments` and `/journals/:id/attachments` return the conventional `{ data: [...] }`. Both credit-note types were unmeasured (every sampled record was void; see below). `listAttachments` in `core/api/attachments.ts` reads either key, so through the CLI or a tool you get `.data` in all cases; only a raw request sees the split. The attachment ID field is `attachmentResourceId` (not `resourceId`). **A VOIDED record 404s on three of the five types.** Measured 2026-09-07: a voided invoice, customer credit note and supplier credit note each answer `404 not_found` with `"<Entity> is voided"`, while a voided bill and a voided journal return 200 and list normally. The 404 is about the record's status, not its existence (`GET /invoices/:id` on the same id works), so never read this 404 as a bad resourceId. Anything fanning out to attachments alongside other calls must treat that leg as degradable, or one voided record destroys the whole response; `get_transaction_summary` did exactly that until 5.55.1.
@@ -222,7 +222,6 @@ The rest of this skill (field names, gotchas, error catalog, dependency order, s
 
 ### Entity Resolution (Fuzzy Matching)
 78. **`--contact`, `--account`, and `--bank-account` accept names**: any CLI flag that takes a contact, chart of accounts entry, or bank account accepts EITHER a UUID resourceId OR a fuzzy name. Examples: `--contact "ACME Corp"`, `--account "DBS Operating"`, `--bank-account "Business"`. The CLI auto-resolves to the best match (strict thresholds) and shows the resolved entity on stderr. UUIDs are passed through without API calls. If the match is ambiguous, the CLI errors with a list of candidates, never silently picks the wrong entity.
-79. **`capsule-transaction` recipes auto-resolve accounts**: when `--input` is omitted, the CLI searches the org's chart of accounts for each blueprint account name (e.g., "Interest Expense", "Loan Payable"). If all accounts resolve with high confidence, no JSON mapping file is needed. If any fail, the error message shows exactly which accounts could not be found and suggests close matches. `--contact` and `--bank-account` on recipes also accept names.
 80. **Payment/refund account is gated by `paymentMethod`, not by the field's type**: the field itself accepts any chart-of-accounts id (the REST layer applies no type filter), but the platform then rejects anything outside Bank Accounts / Cash **when the method is BANK_TRANSFER, CASH or CHEQUE**: `INVALID_ACCOUNT_FOR_BUSINESS_TRANSACTION_FOUND`. Every other method reaches a non-bank account, which is the only way a deposit drawdown works: pay an invoice with `paymentMethod: OTHER` and `accountResourceId` set to the deposit account. The CLI mirrors this, filtering `--account` to bank/cash for those three methods only. Do not read the permissive field as permission: the REST layer is a proxy, and the accounting engine behind it is the authority.
 
 ### Draft Finalization Pipeline (Convert & Next)
@@ -454,48 +453,16 @@ Bills, invoices, and credit notes share identical mandatory field specs. Adding
 
 141. **`bulk_upsert_chart_of_accounts`: sync bulk-upsert with PARTIAL_SUCCESS**. Wraps `POST /api/v1/chart-of-accounts/bulk-upsert` (max 500 per call). Returns synchronously (no jobId polling): `{ resourceIds: string[], failedRows: ImportedRowError[], failedCount: number }` per rule 136. Each successful row contributes one `resourceId`; each failure surfaces a `failedRows[]` entry with `rowIndex` (1-based per the API), `columnName`, `columnValue`, `errorCode`, `errorMessage`. **Dedup is by NAME, not code**; collisions emit `ORGANIZATION_CHART_OF_ACCOUNT_DUPLICATED` per row (other rows in the batch still succeed). Provide `resourceId` per account to update; omit to create. Accepts the all 23 `accountType` values per rules 21 and 140 (variants normalized client-side). For one-off creates with auto-dedup-on-name (returns existing if found), use `create_account` instead. CLI counterpart: `clio accounts bulk-upsert --input <file.json>`.
 
-142. **`capsuleRecipe` payload is mutually exclusive with `capsuleResourceId`** on trigger mutations (create/update of invoice, bill, journal, cash_in, cash_out). Use `capsuleRecipe` to CREATE a new capsule via the recipe engine; use `capsuleResourceId` to ATTACH a base-trx to an existing capsule. Sending both returns 422 (`excluded_with` validator).
-
-143. **Capsule recipe publish is best-effort post-commit: silent-null failure mode.** On success the trigger-mutation response carries `capsuleRecipeJob: { jobResourceId, capsuleResourceId, subscriptionFBPath, totalRecords, idempotentHit, recipeKey }` (verified live 2026-05-27). **Note `jobResourceId` (NOT `resourceId`)** on the trigger-mutation payload; this is the polling key. **On publish failure, `capsuleRecipeJob` is absent (or null) from the response, the trigger mutation STILL returns its normal success status (201 on create, 200 on update; see endpoints.md "Success Status Codes"), the base-trx is committed, and NO error reason is surfaced to the caller.** The response echoes `capsuleRecipe.{recipeName, inputs}` back unchanged, which can look like success at a glance. **Three known causes** of silent null `capsuleRecipeJob` (must pre-validate before sending):
-  - **(a) Wrong `recipeName` for the base trx type**: every recipe is locked to `allowedBaseTransactionTypes` (e.g. PREPAID_AMORTIZATION = PURCHASE only; DEFERRED_REVENUE = SALE only; ACCRUAL_REVERSAL, IFRS16_LEASE = JOURNAL_MANUAL only; LOAN_AMORTIZATION = JOURNAL_DIRECT_CASH_IN or JOURNAL_MANUAL). On `preview_capsule_recipe`, mismatch surfaces as 422 `RECIPE_INVALID_BASE_TRANSACTION_TYPE`. **On the trigger mutation, it silently nulls the job**; the validation happens post-commit in customer-service and arap catches the exception. Always check `get_capsule_recipe(name).allowedBaseTransactionTypes` matches the trigger mutation you're calling.
-  - **(b) Currency mismatch**: see Rule 156 (single-currency v1 recipes: recipe `currency`, every `*AccountResourceId` account's `currencyCode`, and the base trx currency MUST all match). Mismatch surfaces as 422 `ERR_RECIPE_ACCOUNT_CURRENCY_MISMATCH` on `preview_capsule_recipe` but silently nulls on the trigger mutation.
-  - **(c) Wrong `x-accountClass` on an input field**: see Rule 157 (each `*AccountResourceId` slot has a required account class). Mismatch silently nulls; preview returns the matching `RECIPE_FIELDS_*` 422.
-  
-  **Diagnosis sequence when the response has null/absent `capsuleRecipeJob`:**
-  1. **Re-run `preview_capsule_recipe`** with the same `recipeName` + `inputs` (without a base trx). This is the canonical pre-flight: it returns the exact 422 reason (`ERR_RECIPE_ACCOUNT_CURRENCY_MISMATCH`, `RECIPE_INVALID_BASE_TRANSACTION_TYPE`, `RECIPE_FIELDS_MUST_DIFFER`, etc.). Most reliable diagnostic: fix the input and retry the trigger.
-  2. **Poll `search_background_jobs --filter '{"baseTransactionResourceId":{"eq":"<id>"}}'`**: if a `FAILED` job exists, its `errorDetails` has the publish failure reason. **If no job exists for the base-trx, the publish never queued** (validation rejected pre-queue in customer-service).
-  3. **`resume_capsule_recipe(capsuleResourceId)`** is only available if a capsule WAS created: i.e. the recipe partially ran. For pre-queue rejections (3 causes above), no capsule exists; the only recovery is to re-issue the trigger mutation with corrected inputs.
-  
-  Pre-flight gate (recommended for agents and integrations): always call `preview_capsule_recipe(recipeName, inputs)` before the trigger mutation. Preview is pure-compute (no side effects) and surfaces every input/account/currency problem with a clear error_type, which eliminates the silent-null class entirely. **Same gate covers `templateOverrides`** (Customize Recipe): pass `capsuleRecipe.templateOverrides: [{slotKey, template}]` to customize generated text (capsule title/description, leg labels, line memos, schedule reference). Valid slotKeys + `{{variables}}` come from `get_capsule_recipe → versions[].templateSlots[]`; `slotKey` ≤128 chars (no dups), `template` ≤2000 (empty string clears a nullable slot). Invalid overrides return 422 `ERR_RECIPE_OVERRIDE_*` (UNKNOWN_RECIPE / MISSING_SLOT_KEY / DUPLICATE_SLOT / UNKNOWN_SLOT / NON_NULLABLE_BLANK / TEMPLATE_TOO_LONG / UNKNOWN_VARIABLE) on preview but silently null on the trigger path, so preview first.
-
-144. **`recipeName` IS enum-constrained at the API layer** (verified live 2026-05-27): closed enum `LOAN_AMORTIZATION | ACCRUAL_REVERSAL | PREPAID_AMORTIZATION | DEFERRED_REVENUE | IFRS16_LEASE` on `POST /capsule-recipes/preview` and on `capsuleRecipe.recipeName` payloads on trigger mutations. Send a string not in the set → 422 validation_error. Don't hard-code the 5 values in motherboard descriptions; discover via `list_capsule_recipes` (the source of truth) and pass the discovered name through.
-
 145. **Pseudo-SQL `truncated:true` does NOT mean "you hit the cap"**; it means "more rows matched than were returned in this preview". Inspect `rowCount` vs preview cap (100) or your LIMIT to interpret. If you need every row, switch to `export_pseudo_sql`.
 
 146. **Pseudo-SQL export `downloadUrl` is S3 pre-signed with ~15min expiry** (`X-Amz-Expires=900`). Fetch immediately; don't store the URL. If a fetch returns 403 (expired), call `get_pseudo_sql_export(jobId)` again for a fresh URL.
 
 147. **Cashflow report** (`download_export(exportType='cashflow')`) returns the org's CASHFLOW template (IAS 7). If no template configured, returns 404 `template_not_found`. Configure via Jaz settings before invoking.
 
-148. **`resume_capsule_recipe` after `terminalReason=BLOCKED_AFTER_3_RESUME_ATTEMPTS` is unavailable.** Only path forward is `rollback_capsule_recipe(capsuleResourceId)` or manual cleanup via Jaz admin. Resume is NOT idempotent; each call counts toward the 3-attempt limit.
-
-149. **`rollback_capsule_recipe` returning `status=PARTIAL_ROLLBACK`** with `blockedAtomResourceIds[]` is safe to retry (rollback is idempotent on already-deleted atoms). Persistent partial-rollback typically indicates an atom is referenced downstream; escalate to ops if retry doesn't resolve.
-
-150. **`preview_capsule_recipe` returns 422 `RECIPE_INVALID_BASE_TRANSACTION_TYPE`** when the recipe's `allowedBaseTransactionTypes` doesn't include the supplied base trx type (see descriptor at `get_capsule_recipe`). **The trigger mutation does NOT surface this 422 to the caller**; it silently nulls `capsuleRecipeJob` on the response (see Rule 143). The allowed types are: PREPAID_AMORTIZATION → PURCHASE only; DEFERRED_REVENUE → SALE only; ACCRUAL_REVERSAL → JOURNAL_MANUAL only; IFRS16_LEASE → JOURNAL_MANUAL only; LOAN_AMORTIZATION → JOURNAL_DIRECT_CASH_IN or JOURNAL_MANUAL. Always check `get_capsule_recipe(name).allowedBaseTransactionTypes` before sending `capsuleRecipe` on any trigger mutation.
-
-151. **Sending BOTH `capsuleRecipe` AND `capsuleResourceId`** on the same trigger mutation returns 422 (`excluded_with` validator; same lock as Rule 142). Pick one based on intent.
-
-152. **`saveAsDraft: true` + `capsuleRecipe` payload**: recipe is stashed in the draft's `pending_capsule_recipe` JSONB column and fires on draft activation (not on draft create). The base-trx commits as DRAFT immediately; the recipe job is created later when the draft is activated via `convert_drafts_to_active`.
-
 153. **Pseudo-SQL `Idempotency-Key` dedup is server-side primary key**: same key + DIFFERENT query body returns the prior job's result (the server does NOT cross-check the new query body). For agent reliability, `run_pseudo_sql_and_download` auto-keys from `sha256(query).slice(0,16)` so dedup is query-tied automatically. If you call `export_pseudo_sql` directly with a manual key, treat it as a per-intent token; don't reuse across different queries.
 
-154. **`rollback_capsule_recipe` on a non-recipe capsule** (a capsule created by the legacy `create_capsule` tool or imported, not by the recipe engine) returns 422 `RECIPE_ROLLBACK_JOB_NOT_FOUND` ("No CAPSULE_RECIPE job found for capsule X in organization Y — nothing to roll back"). Rollback only works on capsules whose lifecycle was managed by the recipe engine. For legacy capsules, use `delete_capsule` instead.
-
 155. **Pseudo-SQL schema is canonical; call `get_pseudo_sql_schema` before any query.** The response returns the live curated catalog (tables / columns / joins / functions) PLUS the canonical `jaz-pseudo-sql.md` skill body in `agentSkillsDoc.content`. Drop the `.md` body into context as the syntax-rules source; treat `tables[] / joins[] / functions[]` as the column-list source. **Cache contract:** the `version` field is a stable 16-char hex hash; within a session, cache by version and don't re-call unless you have reason to believe the schema changed (e.g. a fresh backend deploy mid-session). Don't re-fetch on a wall-clock timer (upstream is `private, no-cache, must-revalidate`). A static curated-schema snapshot used to ship with the `jaz-pseudo-sql` skill before v5.6.0; it was dropped because it was structurally guaranteed to drift. Never write a query from a memorized column list.
 
-156. **v1 capsule recipes are single-currency: `ERR_RECIPE_ACCOUNT_CURRENCY_MISMATCH`.** The recipe `currency` field, every `*AccountResourceId` account's `currencyCode`, and the base transaction's `currencyCode` ALL MUST match. Preview returns 422 `ERR_RECIPE_ACCOUNT_CURRENCY_MISMATCH` with a concrete message (e.g. "account X is denominated in USD but the recipe currency is SGD"). **Trigger mutation silently nulls `capsuleRecipeJob`** (Rule 143). Practical recipe: never hardcode `currency`; derive from the account's `currencyCode` (`search_accounts(filter: {resourceId: {eq: <prepaidAssetAccountResourceId>}})`) (or the base trx currency) and use the same value as the recipe input. The recipe descriptor's `currency.x-baseTrxBinding: "strict"` field marks this: when present, the value is bound to (and must equal) `trx.currencyCode` post-commit. Caught by smoke runs since v5.5.0; tests 65/66 hardcoded `SGD` and silently nulled against a USD fire-test org for 20+ hours.
-
-157. **Recipe input `*AccountResourceId` fields are account-class-locked; `x-accountClass` in inputSchema is authoritative.** Each `*AccountResourceId` slot on a recipe's input schema carries an `x-accountClass` constraint (`"Asset"`, `"Liability"`, `"Expense"`, `"Revenue"`, `"Equity"`). Passing an account whose class doesn't match the slot's `x-accountClass` is rejected post-commit and silently nulls `capsuleRecipeJob` (Rule 143). Schema location: `get_capsule_recipe(name).data.versions[0].inputSchema.properties.<fieldName>['x-accountClass']`; **note `versions[0].inputSchema`, NOT `inputSchema` at the top level**. Examples: PREPAID_AMORTIZATION needs `prepaidAssetAccountResourceId: Asset` + `expenseAccountResourceId: Expense`; DEFERRED_REVENUE needs `deferredRevenueAccountResourceId: Liability` + `revenueAccountResourceId: Revenue`; ACCRUAL_REVERSAL needs `expenseAccountResourceId: Expense` + `accruedLiabilityAccountResourceId: Liability`. Always pre-validate the account's `accountClass` (`search_accounts(filter: {resourceId: {eq: <id>}})`) against the slot constraint, or just call `preview_capsule_recipe(recipeName, inputs)` to surface every class violation as a clean 422.
-
 ## Supporting Files
 
 For detailed reference, read these files in this skill directory:
@@ -559,4 +526,4 @@ When the user wants to OPEN, SEE, or SHARE something in the Jaz dashboard ("open
 
 160. **Payment `adjustment` is a CASH-LEG-only correction; it never moves the document balance.** Records an overpayment or a rounding difference on the bank side of a payment: `netCash = paymentAmount -/+ fees +/- adjustmentValue` (fees are deducted from cash received on an invoice and added to cash spent on a bill). AR/AP is untouched, so an overpaid invoice stays PAID with the excess sitting on the account you chose and NO credit note is created. Shape: `adjustment: { adjustmentValue, adjustmentAccountResourceId, adjustmentDescription? }`, signed, non-zero, max 2dp, always FLAT (never a percentage), never taxed. Accepted on `pay_invoice` / `pay_bill` / both credit-note refunds / `update_payment` / both receipt reconciliations. **NOT** on batch payments, and not for `DEBT_WRITE_OFF` / `CLEARING_SETTLEMENT` / `INTER_COMPANY` / `WITHHOLDING_TAX_CERTIFICATE` (those record a settlement, so there is no cash leg). The account must be non-controlled: not AR/AP, not the VAT or FX accounts, not bank/cash, not deposit-linked. **Write/read asymmetry:** you WRITE the nested `adjustment.adjustmentValue` but READ a flat `adjustmentAmount` on the payment record, alongside `adjustmentOrganizationAccountResourceId`. Don't confuse either with `lineItems[].taxVatAdjustment.adjustmentAmount`, which is line-item tax and unrelated. **On `update_payment`, adding an adjustment and changing an existing one both APPLY.** Verified against production on arap `v10.7.21` (2026-08-11): an amount move and an account-plus-description move both returned 200 and applied. `paymentAmount` and `transactionAmount` are untouched. **Removing one does NOT work.** `adjustmentValue: 0` is rejected at the edge with a 422, and `"adjustment": null` is indistinguishable from omitting the field, so both leave the stored value unchanged. To remove an adjustment, delete the payment record and record it again without one. An explicit clear flag is being added upstream; zero stays rejected, so omit-to-clear will never be the answer. Anything describing this field as inert on update is stale. **arap `v10.7.21` put both guards behind one comparator, and a CHANGE now has two rejection conditions.** `PaymentAmountCalculator.adjustmentChanged` answers "did the adjustment change?" for the reconciled check AND for lock-date validation, which previously disagreed: the reconciled check compared the amount alone, and lock-date validation did not look at the adjustment at all. It compares **amount, account AND description** (amount by `compareTo`, so `5.0` and `5.00` are equal). Two conditions reject a change: **(a) the payment is RECONCILED** to a bank statement entry, giving `PAYMENT_ADJUSTMENT_CANNOT_CHANGE_WHEN_RECONCILED` (moving it to a different account trips this now, where the amount-only check let it through); **(b) an account on the payment's LEDGER ROWS carries a lock date later than the `valueDate`**, giving `DELETE_TRANSACTION_VALUE_DATE_CANNOT_BE_EARLIER_THAN_LOCK_DATE` -- an update deletes and recreates the ledger rows, so the delete half trips first and the code names a delete for a call that was an update. **The escape hatch:** resending the same adjustment, or omitting the field, compares equal and is NOT a change, so it does not by itself trip either guard. **But that is not a general exemption, and this is the trap.** `adjustmentDescription` is optional, so resending only `adjustmentValue` + `adjustmentAccountResourceId` against a payment that HAS a description compares unequal and counts as a change. Resend all three, or omit the object entirely. And lock-date validation gates on seven other fields independently: changing `reference`, `valueDate`, `paymentMethod`, `organizationAccountResourceId`, `paymentAmount`, `transactionAmount`, or converting a draft to active is lock-date validated on its own. Only edits that move no ledger row survive a lock. **Measured on production 2026-08-11**, throwaway account locked at 2026-07-31 over a payment dated 2026-06-15: change the adjustment -> 422; change ONLY `reference` -> 422; resend all three values unchanged -> 200; resend WITHOUT the description -> 422; then unlock and repeat the first call -> 200 and applied. That last row is the control. Condition (a) is read from the deployed source, not triggered: it needs a bank statement entry, which has no delete, so proving it costs permanent state on a shared org.
 
-161. **Report templates: the saved layout of a report, one default per report type, and packs.** Eight verbs on MCP and CLI (list, get, search, default-configuration, create, update, set-default, delete), each taking a template by **resourceId or exact name**; render one with `download_export` / `clio exports download --template <name>`. `templateConfiguration` is a JSON-encoded string on the wire; Clio hands it over as an object, never shows the server-owned `coaSnapshot` (a copy of the whole chart of accounts, up to 324 KB), and returns a large layout as a key map plus a row outline. Change part of one with `edits` (set / insert / remove by path; updateRow / addRow / removeRow / moveRow by row id), applied all or nothing and refused when they introduce a silent fault (an unknown enum value, a formula naming a missing row, a removed engine total). **Set default changes every export and dashboard view** of that report for the whole organization; the answer carries `previousDefault` to undo it. A pack's reports come from `packTemplates` (names or ids, in order), written to both places the server reads them from. Measured upstream quirks (2026-09-24): an export given a template it cannot use (unknown, or of another report type) answers **HTTP 200 with an empty body**, so Clio checks the template first; for P&L and cashflow only `IAS_1` and `IFRS_18` have a default layout (`GAAP` / `IND_AS` / `IFRS_SME` answer 422 `NO_TEMPLATE_FOUND` without one); profit-and-loss and balance-sheet saves drop keys the server does not model with no error (Clio reports them as `notSaved`); deleting a pack's last report leaves an empty pack whenever the pack has any other component (Clio refuses it); four of five search filters answer 500 (Clio filters the list). Full reference: `references/report-templates.md`.
+161. **Report templates: the saved layout of a report, one default per report type, and packs.** Eight verbs on MCP and CLI (list, get, search, default-configuration, create, update, set-default, delete), each taking a template by **resourceId or exact name**; render one with `download_export` / `clio exports download --template <name>`. `templateConfiguration` is a JSON-encoded string on the wire; Clio hands it over as an object, never shows the server-owned `coaSnapshot` (a copy of the whole chart of accounts, up to 324 KB), and returns a large layout as a key map plus a row outline. Change part of one with `edits` (set / insert / remove by path; updateRow / addRow / removeRow / moveRow by row id), applied all or nothing and refused when they introduce a silent fault (an unknown enum value, a formula naming a missing row, a removed engine total). **Set default changes every export and dashboard view** of that report for the whole organization; the answer carries `previousDefault` to undo it. A pack's reports come from `packTemplates` (names or ids, in order), written to both places the server reads them from. Measured upstream quirks (2026-09-24): an export given a template it cannot use (unknown, or of another report type) answers **HTTP 200 with an empty body**, so Clio checks the template first; for P&L and cashflow only `IAS_1` and `IFRS_18` have a default layout (`GAAP` / `IND_AS` / `IFRS_SME` answer 422 `NO_TEMPLATE_FOUND` without one); profit-and-loss and balance-sheet saves drop keys the server does not model with no error (Clio reports them as `notSaved`); deleting a pack's last report leaves an empty pack whenever the pack has any other component (Clio refuses it); upstream search refuses `and` / `or` and name sort (Clio filters the list). Full reference: `references/report-templates.md`.

package/assets/skills/api/references/endpoints.md

@@ -2927,7 +2927,7 @@ Every template of the organization in one response, **under `reportTemplates`, n
 
 ### POST /api/v1/organization-report-template/search
 
-`{ filter, sort }`, answers a bare array. **Four of its five filters answer 500** (`reportCategory` without `reportType`, `isDefault`, `reportTypes`, `resourceId`), and there is no name filter. Clio filters the list instead.
+`{ filter, sort }`, answers a bare array. Honours `isDefault`, `reportCategory`, `resourceId`, `reportType` (plain string), `reportTypes` (plain array) and `templateName` (eq, contains); refuses `and` / `or` (400) and sorting by `templateName` (422). Measured 2026-10-01. Clio filters the full list instead, which supports all of them.
 
 ### GET /api/v1/organization-report-template/default-configuration
 

package/assets/skills/api/references/full-api-surface.md

@@ -521,7 +521,7 @@ Body for all three: `{ items: [{btResourceId: "<uuid>", btType: "SALE|PURCHASE|S
 |--------|------|-------------|
 | GET | `/organization-report-template` | List every template (rows under `reportTemplates`; ignores paging) |
 | GET | `/organization-report-template/:resourceId` | Get one |
-| POST | `/organization-report-template/search` | Search (4 of 5 filters answer 500; Clio filters the list instead) |
+| POST | `/organization-report-template/search` | Search (no `and` / `or` or name sort; Clio filters the list instead) |
 | GET | `/organization-report-template/default-configuration` | The default layout a new template of a report type starts from |
 | POST | `/organization-report-template` | Create (layout defaults to the system layout; a pack needs its layout and `packTemplates`) |
 | PUT | `/organization-report-template/:resourceId` | Rename, replace the layout, or replace a pack's reports |

package/assets/skills/api/references/report-templates.md

@@ -289,7 +289,7 @@ clio report-templates create --name "Board pack" --type REPORT_PACK \
 
 ### Search
 
-Upstream's search endpoint answers 500 for four of its five filters, and has no name filter, so `search_organization_report_templates` filters the full list instead. That matches because the list is every template at once. It uses the same filter grammar, plus `templateName`:
+Upstream's search endpoint now honours `isDefault`, `reportCategory`, `resourceId`, `reportType`, `reportTypes` and `templateName` (eq, contains), but still refuses `and` / `or` and sorting by `templateName` (measured 2026-10-01), so `search_organization_report_templates` filters the full list instead. That matches because the list is every template at once. It uses the same filter grammar, plus `templateName`:
 
 ```jsonc
 { "filter": { "or": [ { "templateName": { "contains": "board" } }, { "isDefault": { "eq": true }, "reportType": "REPORT_PACK" } ] } }

package/assets/skills/api/references/search-enums.md

@@ -548,7 +548,7 @@ All search endpoints return:
 }
 ```
 
-**Exception**: `GET /organization-report-template` returns its rows under a `reportTemplates` key, not `data` (measured 2026-09-06). Its `/search` sibling returns a **bare array** with no envelope (measured 2026-09-07). The tools normalize the list; read `.data` on both. Since 2026-09-24 the report template search tool filters the list itself instead of calling `/search`, because four of the search endpoint's five filters answer 500; it takes the same filter grammar plus `templateName` (`references/report-templates.md`).
+**Exception**: `GET /organization-report-template` returns its rows under a `reportTemplates` key, not `data` (measured 2026-09-06). Its `/search` sibling returns a **bare array** with no envelope (measured 2026-09-07). The tools normalize the list; read `.data` on both. Since 2026-09-24 the report template search tool filters the list itself instead of calling `/search`, because the search endpoint still refuses `and` / `or` and sorting by name (measured 2026-10-01); it takes the same filter grammar plus `templateName` (`references/report-templates.md`).
 
 ---
 

package/assets/skills/api/references/search-reference.md

@@ -28,7 +28,7 @@ All search endpoints share this identical structure:
 
 **Response shape**: `{ totalElements, totalPages, data: [...] }`; all search/list endpoints return this flat structure directly (no outer `data` wrapper).
 
-> **Exception**: the two halves of this family answer differently upstream. `POST /organization-report-template/search` returns a **bare array** with no envelope; `GET /organization-report-template` returns its rows under a **`reportTemplates`** key with no counts. Measured 2026-09-07: the two halves disagree with each other and with the house style. The tools normalize the list, so you always read `.data`. The report template search tool does not call `/search` at all: four of its five filters answer 500 upstream (re-measured 2026-09-24), so it filters the full list with the same grammar plus `templateName`. See `references/report-templates.md`.
+> **Exception**: the two halves of this family answer differently upstream. `POST /organization-report-template/search` returns a **bare array** with no envelope; `GET /organization-report-template` returns its rows under a **`reportTemplates`** key with no counts. Measured 2026-09-07: the two halves disagree with each other and with the house style. The tools normalize the list, so you always read `.data`. The report template search tool does not call `/search` at all: upstream still refuses `and` / `or` and sorting by name (measured 2026-10-01), so it filters the full list with the same grammar plus `templateName`. See `references/report-templates.md`.
 
 ---
 

package/assets/skills/cli/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: jaz-cli
-version: 5.74.1
+version: 6.0.0
 description: >-
   Use this skill when running Clio CLI commands, building shell scripts with
   Clio, debugging auth issues, understanding --json output, paginating results,
@@ -8,7 +8,7 @@ description: >-
   precedence, output formats, entity resolution, and common workflow patterns.
   Also use when the user asks how to use clio, what commands are available, or
   how to automate accounting tasks from the command line. Covers all
-  76 command groups and 386 tools, including employee-expense claims.
+  74 command groups and 381 tools, including employee-expense claims.
 license: MIT
 compatibility: Requires Node.js >= 18.0.0. Install via npm install -g jaz-clio.
 ---
@@ -17,7 +17,7 @@ compatibility: Requires Node.js >= 18.0.0. Install via npm install -g jaz-clio.
 
 > **Audience note:** for power users and CI/automation. Load this skill only when you're scripting from a terminal, building shell pipelines, or debugging from `clio --json` output. For day-to-day accounting inside Claude Desktop / Cowork, the MCP tools cover the common flows without dropping to the CLI.
 
-You are working with **Clio** (`jaz-clio`), the CLI for the Jaz accounting platform. 76 command groups, 13 calculators, 12 job blueprints, 386 tools. Also fully compatible with Juan Accounting (same API, same endpoints).
+You are working with **Clio** (`jaz-clio`), the CLI for the Jaz accounting platform. 74 command groups, 13 calculators, 12 job playbooks (jaz-jobs skill), 381 tools. Also fully compatible with Juan Accounting (same API, same endpoints).
 
 ## When to Use This Skill
 
@@ -180,7 +180,6 @@ Rules:
 | `--ref <reference>` | Search/create commands | Reference string |
 | `--tag <name>` | Search/create commands | Tag filter or assignment |
 | `--input <file>` | Create/update commands | Read full JSON body from file |
-| `--plan` | Recipe commands | Offline plan mode (no auth) |
 
 ## Search Query Expressions (`--query`)
 
@@ -277,13 +276,13 @@ Sending `lineItems[]` to the FLAT endpoint silently ignores them and creates a $
 
 **Reports & Exports**: `reports` (16 report types), `exports`
 
-**AI & Automation**: `magic` (create, status), `quick-fix`, `ledger-find-fix` (preview, apply), `capsules`, `capsule-transaction` (alias: `ct`, 13 recipe types)
+**AI & Automation**: `magic` (create, status), `quick-fix`, `ledger-find-fix` (preview, apply), `capsules`
 
 **Judgment journal**: `jots` (create, recall, dispose). Every write command also takes `--jot "<one line>"` to log the judgment behind that specific write inline, no extra call (see `--jot` in Common Flags).
 
 **Calculators**: `calc` (loan, lease, depreciation, prepaid-expense, deferred-revenue, fx-reval, ecl, provision, fixed-deposit, asset-disposal, accrued-expense, leave-accrual, dividend)
 
-**Jobs**: `jobs` (month-end, quarter-end, year-end, bank-recon, gst-vat, payment-run, credit-control, supplier-recon, audit-prep, fa-review, document-collection, statutory-filing) + tools (match, outstanding, ingest, sg-cs, sg-ca)
+**Jobs**: `jobs` (`bank-recon match`, `payment-run outstanding`, `document-collection ingest`, `statutory-filing sg-cs`, `statutory-filing sg-ca`). The job playbooks themselves are in the **jaz-jobs** skill.
 
 **Organization**: `org` (info), `org-users`, `auth`
 
@@ -295,7 +294,7 @@ See `references/command-catalog.md` for the full catalog with subcommands and fl
 
 ## Offline vs Online
 
-Offline commands (no auth needed): `calc`, `jobs` (blueprints only), `capsule-transaction --plan`, `help-center`, `init`, `versions`, `update`
+Offline commands (no auth needed): `calc`, `jobs bank-recon match`, `jobs statutory-filing sg-cs` / `sg-ca`, `help-center`, `init`, `versions`, `update`
 
 Everything else requires authentication (API key).
 
@@ -357,32 +356,23 @@ Transaction create commands (`invoices`, `bills`, `customer-credit-notes`, `supp
 
 This catches mistakes before the API call, saving round-trip time and providing clearer error messages.
 
-## Capsule-Transaction Recipes
+## Calculated Schedules (calculate, capsule, post)
 
-The `capsule-transaction` (alias: `ct`) command group is the most powerful CLI feature. Each subcommand:
+There is no one-shot recipe command. A loan, lease, prepaid, deferred revenue, accrual, ECL, provision, fixed deposit, disposal or dividend is three steps:
 
-1. Runs a financial calculator (same as `clio calc`)
-2. Creates a capsule to group the transactions
-3. Posts all transactions (invoices, bills, journals) in sequence
-
-Two entry paths:
-- **Full**: provide `--input` (account mapping) or let it auto-resolve from your chart of accounts
-- **Attach**: provide `--existing-txn <id>` to skip the initial transaction and create only the delta (e.g., attach depreciation to an existing purchase bill)
-
-Plan mode (`--plan`) is offline and shows what accounts are needed and what steps will be created, without making any API calls.
+1. **Calculate (offline):** `clio calc <type> ... --json` returns the schedule and, when `--start-date` is given, a `blueprint`: the capsule type and name plus dated steps with journal lines. It posts nothing.
+2. **Create the capsule:** `clio capsules types` to find the type id (`clio capsules create-type --name <name>` if it is missing), then `clio capsules create --type <typeId> --title <title>`.
+3. **Post each step** with `clio journals create` / `clio bills create` / `clio invoices create` / `clio cash-in create` / `clio cash-out create`, passing a JSON body via `--input` that includes `capsuleResourceId`. A fixed amount repeating every period can be one `clio schedulers create-journal`.
 
 ```bash
-# Plan mode: see what's needed (offline)
-clio ct loan --principal 100000 --rate 5 --term 60 --plan
-
-# Execute with auto-resolve (uses fuzzy matching against your chart of accounts)
-clio ct loan --principal 100000 --rate 5 --term 60 --start-date 2026-01-01 --ref LOAN-001
-
-# Execute with explicit account mapping
-clio ct loan --principal 100000 --rate 5 --term 60 --start-date 2026-01-01 \
-  --input account-mapping.json --bank-account "Bank - SGD" --contact "HSBC"
+clio calc loan --principal 100000 --rate 5 --term 60 --start-date 2026-01-01 --json > loan.json
+clio capsules types --json
+clio capsules create --type <typeId> --title "Bank loan 2026" --json
+clio journals create --input step-2.json --json   # body includes capsuleResourceId
 ```
 
+`clio calc fx-reval` is verification only: Jaz revalues foreign-currency balances at period end, so do not post its result.
+
 ## Tips
 
 1. **Pipe JSON to jq**: `clio invoices list --json | jq '.data[] | {ref: .reference, amount: .totalAmount}'`
@@ -391,7 +381,7 @@ clio ct loan --principal 100000 --rate 5 --term 60 --start-date 2026-01-01 \
 4. **Draft-then-finalize**: The CLI defaults to saving as draft (overrides the API default of `saveAsDraft: false`). Use `--finalize` to create a finalized transaction immediately. Note: `cash-in`, `cash-out` and `cash-transfer` have no draft state and so take no `--finalize`; they always post ACTIVE.
 5. **Idempotent creates**: Use `--input` with the same JSON to get consistent results. The API dedup guards catch duplicate contacts, items, and accounts.
 6. **Check before bulk ops**: Always preview with `--json | jq length` before piping IDs into `quick-fix`. To find & fix (recode, retag, re-date) records found by a condition, across record types, `clio ledger-find-fix preview` lists what would change and why anything is excluded, and `clio ledger-find-fix apply -- <previewId>` runs exactly that, once.
-7. **Offline calculators for exploration**: `clio calc` commands need no auth -- use them to explore scenarios before committing with `clio ct`.
+7. **Offline calculators for exploration**: `clio calc` commands need no auth -- use them to explore scenarios before posting anything.
 8. **Help center for guidance**: `clio hc "how to reconcile"` searches the full Jaz help center locally (hybrid: embeddings + keyword).
 
 See `references/common-workflows.md` for end-to-end multi-command patterns.

package/assets/skills/cli/references/agent-gotchas.md

@@ -27,9 +27,9 @@
 
 11. **`clio schema` is for tool introspection, not data.** `clio schema --json` lists command groups and tool counts. `clio schema invoices --json` shows tool definitions. `clio schema invoices create --json` shows parameter schema. None of these hit the API.
 
-12. **Job blueprints output plain text by default.** `clio jobs month-end` prints a checklist. Add `--json` for structured output. Some job tools (`match`, `outstanding`, `sg-cs`) require auth; blueprint generators are offline.
+12. **`clio jobs` is five working tools, not checklists.** `bank-recon match`, `payment-run outstanding`, `document-collection ingest`, `statutory-filing sg-cs` and `statutory-filing sg-ca`. No checklist commands exist (month-end, GST/VAT, audit prep and the rest are not commands): the job playbooks are reference docs in the jaz-jobs skill. `payment-run outstanding` and `ingest --upload` need auth; `match`, `sg-cs` and `sg-ca` run offline. Add `--json` for structured output.
 
-13. **Capsule-transaction recipes: always `--plan` first.** `clio ct loan --plan` shows what accounts and transactions will be created, offline. Only run without `--plan` when you know the account mapping is correct. Recipes create real finalized transactions.
+13. **There is no recipe command: calculate, then post.** `clio calc <type> --json` returns the schedule and journal lines and posts nothing. To book it, create a capsule (`clio capsules create`) and post each step yourself (`journals` / `bills` / `invoices` / `cash-in` / `cash-out` `create`, with `capsuleResourceId` in the `--input` body). The calculator's account names are generic ("Cash / Bank Account"): map each to a real account before posting. `clio calc fx-reval` is verification only; never post its result.
 
 14. **The `--input` flag reads JSON from a file.** For complex payloads (multi-line-item invoices, detailed journals), write JSON to a temp file and pass `--input payload.json` instead of long `--lines` flags with shell escaping issues. When `--input` is provided, all other body flags are ignored.
 

package/assets/skills/cli/references/command-catalog.md

@@ -1,6 +1,6 @@
 # Clio Command Catalog
 
-Complete reference for all 76 command groups. Organized by domain.
+Complete reference for all 74 command groups. Organized by domain.
 
 ---
 
@@ -463,7 +463,7 @@ clio exports download --type profit-and-loss --format PDF --start-date 2026-01-0
 clio report-templates set-default "Board P&L"
 ```
 
-`list` and `search` return summaries without layouts (`configurationChars` gives each layout's size); `get` returns the layout as an object without the server-owned `coaSnapshot`. Search runs over the full list, so every filter works, `--name` included (upstream's own search answers 500 for four of its five filters). A write prints what the server did not keep, if anything (`notSaved`): profit-and-loss and balance-sheet layouts drop keys the server does not model.
+`list` and `search` return summaries without layouts (`configurationChars` gives each layout's size); `get` returns the layout as an object without the server-owned `coaSnapshot`. Search runs over the full list, so every filter works, `--name` included (upstream's own search refuses `and` / `or` and sorting by name). A write prints what the server did not keep, if anything (`notSaved`): profit-and-loss and balance-sheet layouts drop keys the server does not model.
 
 ### `clio pseudo-sql` (alias: `sql`): Read-only SQL over the curated reporting tables (6 tools)
 A restricted SQL subset against Jaz's curated reporting schema, **not** the customer's database. SELECT only, single statement, 16384-char cap; the engine rejects DML, so there is no write verb.
@@ -605,15 +605,11 @@ Every write after `record` takes `--hash` (from `search` or the last write). A l
 | `create` | `--title`, `--type`, `--input` |
 | `update <id>` | `--title`, `--input` |
 | `delete <id>` | |
+| `types` | `--json` (list capsule types; gives the id `create --type` needs) |
+| `create-type` | `--name`, `--description`, `--input` |
+| `move` | `--transactions`, `--to`, `--from` |
 
-### `clio capsule-transaction` (alias: `ct`): Transaction recipes
-13 IFRS-compliant recipe subcommands. Each runs a calculator, creates a capsule, and posts all transactions.
-
-Subcommands: `loan`, `lease`, `depreciation`, `prepaid-expense`, `deferred-revenue`, `fx-reval`, `ecl`, `provision`, `fixed-deposit`, `asset-disposal`, `accrued-expense`, `leave-accrual`, `dividend`
-
-Shared flags: `--plan`, `--input`, `--bank-account`, `--contact`, `--existing-txn`, `--ref`, `--finalize`, `--json`
-
-Each subcommand has calculator-specific required options (e.g., `--principal`, `--rate`, `--term` for loan).
+To post a calculated schedule: `clio calc <type> --json` for the steps, `clio capsules create` for the capsule, then one `journals` / `bills` / `invoices` / `cash-in` / `cash-out` `create` per step with `capsuleResourceId` in the `--input` body.
 
 ---
 
@@ -642,30 +638,18 @@ Common optional flags: `--start-date`, `--currency`, `--json`
 
 ---
 
-## Jobs (Offline Blueprints + Online Tools)
+## Jobs (Working Tools)
 
-### `clio jobs`: 12 job blueprints + tools
-Blueprints are offline (no auth). Tools require auth.
+### `clio jobs`: job tools
+Five working sub-tools. The job playbooks themselves (month-end close, GST/VAT filing, audit prep, ...) are reference docs in the jaz-jobs skill, not commands.
 
-| Subcommand | Type | Key flags |
+| Subcommand | Auth | Key flags |
 |------------|------|-----------|
-| `month-end` | Blueprint | `--period`, `--currency` |
-| `quarter-end` | `--period`, Blueprint | `--currency` |
-| `year-end` | `--period`, Blueprint | `--currency` |
-| `bank-recon` | Blueprint | `--account`, `--period`, `--currency` |
-| `match` | Tool | `--input` (bank records + transactions JSON) |
-| `gst-vat` | Blueprint | `--period` |
-| `payment-run` | Blueprint | `--due-before`, `--currency` |
-| `outstanding` | Tool | `--limit` (group outstanding bills) |
-| `credit-control` | `--overdue-days`, Blueprint | |
-| `supplier-recon` | `--supplier`, `--period`, Blueprint | |
-| `audit-prep` | `--period`, Blueprint | |
-| `fa-review` | `--currency`, Blueprint | |
-| `document-collection` | Blueprint | `--currency` |
-| `ingest` | `--source`, Tool | `<path>` (classify + upload documents) |
-| `statutory-filing` | Blueprint | `--ya`, `--jurisdiction`, `--currency` |
-| `sg-cs` | Tool | `--input` (compute Singapore Form C-S) |
-| `sg-ca` | Tool | `--input` (compute Singapore capital allowances) |
+| `bank-recon match` | Offline | `--input` (bank records + transactions JSON), `--tolerance`, `--date-window`, `--max-group`, `--currency`, `--find-all`, `--json` |
+| `payment-run outstanding` | Required | `--due-before`, `--currency`, `--supplier`, `--limit`, `--json` (outstanding bills grouped by supplier) |
+| `document-collection ingest` | Only with `--upload` | `--source` (local dir, .zip, or share URL), `--type`, `--upload`, `--bank-account`, `--timeout`, `--json` |
+| `statutory-filing sg-cs` | Offline | `--input`, or `--ya`, `--revenue`, `--profit`, `--depreciation`, `--exemption`, `--json` (Singapore Form C-S) |
+| `statutory-filing sg-ca` | Offline | `--input`, or `--ya`, `--cost`, `--category`, `--acquired`, `--json` (Singapore capital allowances) |
 
 ---
 
@@ -750,7 +734,7 @@ Valid entity types: INVOICE, BILL, CUSTOMER_CREDIT_NOTE, SUPPLIER_CREDIT_NOTE, S
 Universal async tracker: any operation returning a jobId (contacts bulk-upsert, items bulk-upsert, bank import, magic processing) can be polled here.
 
 ### `clio mcp`: MCP stdio server
-Starts an MCP server for Claude Code / AI tool integration. Exposes all 386 operations.
+Starts an MCP server for Claude Code / AI tool integration. Exposes all 381 operations.
 
 ### `clio serve`: HTTP daemon
 Starts the HTTP daemon for ChatKit and email channel integrations.

package/assets/skills/cli/references/common-workflows.md

@@ -63,12 +63,9 @@ clio bank records "$BANK_ID" --status POSSIBLE_DUPLICATE --json
 
 ## 3. Month-End Close Workflow
 
-Generate a blueprint, run reports, and create any adjusting journals.
+Follow the month-end playbook in the jaz-jobs skill (`references/month-end-close.md`), run reports, and create any adjusting journals.
 
 ```bash
-# Generate month-end checklist (offline, no auth)
-clio jobs month-end --period 2026-03 --currency SGD --json > month-end-checklist.json
-
 # Run key reports for review
 clio reports generate trial-balance --to 2026-03-31 --json > tb-mar.json
 clio reports generate profit-loss --from 2026-03-01 --to 2026-03-31 --json > pl-mar.json
@@ -134,42 +131,36 @@ clio invoices pay "$INVOICE_ID" \
 
 ## 5. Equipment Purchase with Depreciation
 
-Record a fixed asset purchase, calculate depreciation, and create journal entries via capsule-transaction.
+Register a fixed asset, cross-check its depreciation schedule, and (for a declining-balance method only) post the charges into a capsule.
 
 ```bash
-# Create the fixed asset record
+# Register the asset. Jaz posts straight-line depreciation itself: do not also journal it.
 clio fixed-assets create \
   --name "Office Printer" \
   --type "Office Equipment" \
   --amount 3600 \
   --date 2026-01-01 \
-  --input '{"depreciationMethod":"STRAIGHT_LINE","usefulLifeMonths":36,"salvageValue":0}'
+  --depreciation-start 2026-01-01 \
+  --asset-account "$ASSET_ACCOUNT_ID" \
+  --depreciation-method STRAIGHT_LINE \
+  --effective-life 36 \
+  --residual 0
 
-# Preview the depreciation schedule (offline, no auth)
+# Cross-check the schedule (offline, no auth). --life is in YEARS; --method is sl, ddb or 150db.
 clio calc depreciation \
   --cost 3600 \
   --salvage 0 \
-  --life 36 \
-  --method straight-line \
+  --life 3 \
+  --method sl \
+  --frequency monthly \
   --currency SGD \
   --json
 
-# Execute as a capsule-transaction (creates capsule + monthly journals)
-clio ct depreciation \
-  --cost 3600 \
-  --salvage 0 \
-  --life 36 \
-  --method straight-line \
-  --ref "DEP-PRINTER" \
-  --json
-
-# Or plan first (offline) to see what accounts are needed
-clio ct depreciation \
-  --cost 3600 \
-  --salvage 0 \
-  --life 36 \
-  --method straight-line \
-  --plan
+# Declining balance (ddb / 150db) is not a register method: register the asset with
+# --depreciation-method NO_DEPRECIATION, then post each period's charge from the calculator yourself.
+clio capsules types --json
+clio capsules create --type "$CAPSULE_TYPE_ID" --title "Office Printer depreciation" --json
+clio journals create --input dep-2026-01.json --json   # body: valueDate, journalEntries, capsuleResourceId
 ```
 
 ## 6. Search, Filter, and Bulk-Update Workflow
@@ -244,8 +235,8 @@ eval "$(clio auth shell-init)"
 Ingest documents, extract data via AI, and review results.
 
 ```bash
-# Ingest a folder of mixed PDFs (invoices, bills, bank statements)
-clio jobs document-collection --json
+# Scan and classify a folder of mixed PDFs (invoices, bills, bank statements); add --upload to send them to Jaz
+clio jobs document-collection ingest --source ./client-docs --json
 
 # Or extract a single document (file, URL, or raw HTML)
 clio magic create --file ./invoice-from-supplier.pdf --type bill --json

package/assets/skills/conversion/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: jaz-conversion
-version: 5.74.1
+version: 6.0.0
 description: >-
   Use this skill when migrating accounting data into Jaz, importing from Xero,
   QuickBooks, Sage, MYOB, or Excel exports. Covers the full conversion pipeline:

package/assets/skills/jaz-kit/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: jaz-kit
-version: 5.74.1
+version: 6.0.0
 description: >-
   Use this skill when an accountant, bookkeeper, or owner is running real books
   in Jaz across one or more organizations from the terminal: setting up a

package/assets/skills/jaz-kit/references/flows.md

@@ -7,7 +7,7 @@ Schemas live in `workspace.md` (KIT.md, ORG.md, `.env`) and `close-state.md` (CL
 ## Ground rules for every flow
 
 1. **One organization per session; ORG.md decides which.** Use its `organization_id` on every call. OAuth CLI: `clio <command> --org oauth:<organization_id> --json`. Hosted MCP: pass the same ID explicitly. Never rely on the shared active organization. For the optional API-key route only, source the folder's `.env` in the same command and omit `--org`; verify the key resolves to the recorded ID.
-2. **Draft first.** `saveAsDraft` defaults to `false` in the API; omitting it posts live. Every write in a Jaz Kit flow sets `saveAsDraft: true` (or `--plan` then a non-finalized run for `clio ct`) unless the org's `rules/` explicitly relaxes that transaction type. Finalization happens in the review flow, never as a side effect.
+2. **Draft first.** `saveAsDraft` defaults to `false` in the API; omitting it posts live. Every write in a Jaz Kit flow sets `saveAsDraft: true` unless the org's `rules/` explicitly relaxes that transaction type. Finalization happens in the review flow, never as a side effect.
 3. **Never print a key.** The key lives in the workspace `.env`, which the user pastes into (never chat, never a message, never an error). Source the file; do not read the key value into your own output. `jk-` strings are redacted on sight.
 4. **Judgment gets recorded.** When the user accepts a variance, carries a residual, or overrides a default, call `jot` at that moment. Mechanical steps never jot.
 5. **Report exceptions, not dumps.** Surface the top few items that need a decision; write the full output to the period's `workpapers/`.
@@ -101,7 +101,7 @@ Requires an open session. The playbooks are in the jobs skill; this flow adds st
 
 **3. Open or resume CLOSE.md** at `closes/<period>/CLOSE.md`. Resuming: read it and **do not re-run confirmed steps**. Any step marked `started` but not `confirmed` crashed mid-write; see the resume rule below before touching it.
 
-**4. Walk the playbook**: month-end, quarter-end, or year-end from the jobs skill (`clio jobs <type> --period <period> --json` prints the same phased checklist if you want it as data). For each step:
+**4. Walk the playbook**: month-end, quarter-end, or year-end from the jobs skill (the per-job reference doc is the plan; there is no checklist command). For each step:
 - mark it `started` in CLOSE.md **before** the call that writes
 - create as draft; add each created record to the review queue with a link built by `navigate` (never write a dashboard URL from memory)
 - mark `confirmed` only after the platform confirms

package/assets/skills/jaz-pseudo-sql/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: jaz-pseudo-sql
-version: 5.74.1
+version: 6.0.0
 description: >-
   Use this skill when answering ad-hoc data questions that aren't covered by
   download_export (canonical reports: anomaly, audit, aging, P&L, BS, GL,