@noresponse/mcp 0.6.12 → 0.6.14

package/manifest.json

@@ -40,15 +40,15 @@
       "type": "string"
     }
   },
-  "version": "0.6.12",
+  "version": "0.6.14",
   "tools": [
     {
       "name": "create_assessment",
-      "description": "Assess identifiers for a signup or gateway decision.\n\nAssesses one account creation, login or purchase (flow) and answers with the account's one current risk: account.risk.decision is the decision for this flow (allow, step_up, review or block, where block means your block threshold was reached), and detail holds this assessment's own facts. depth sets how long the answer waits for sources: quick waits only for the fast ones (500 ms), full for the whole budget (3000 ms at most); it defaults to full for account_creation and purchase and quick for login. 200 carries complete or partial, 202 carries pending with Location and Retry-After. While the account's first assessment is pending, account is absent: apply your own fallback for that, 429, 503 or a timeout. NoResponse blocks nothing itself; you decide. Requires the assess scope. Idempotency-Key is optional in the SDK and is bound to the body for 24 hours: the same key with the same body replays the latest assessment with Idempotent-Replayed: true and starts no new work; the same key with a different body answers 409 idempotency_key_reused. response_mode defaults to standard; detailed needs the detailed scope."
+      "description": "Assess identifiers for a signup or gateway decision.\n\nAssesses one account creation, login or purchase (flow) and answers with the account's one current risk: account.risk.decision is the decision for this flow (allow, step_up, review or block, where block means your block threshold was reached), and detail holds this assessment's own facts. depth sets how long the answer waits for sources: quick waits only for the fast ones (500 ms), full for the whole budget (3000 ms at most); it defaults to full for account_creation and purchase and quick for login. 200 carries complete or partial, 202 carries pending with Location and Retry-After. While the account's first assessment is pending, account is absent: apply your own fallback for that, 429, 503 or a timeout. NoResponse blocks nothing itself; you decide. Requires the assess scope. Idempotency-Key is optional in the SDK and is bound to the body for 24 hours: the same key with the same body replays the latest assessment with Idempotent-Replayed: true and starts no new work; the same key with a different body answers 409 idempotency_key_reused. response_mode defaults to standard; detailed needs the detailed scope. The answer leaves out detail.identifiers: you hold what you sent, and a key with the history scope gets them back with GET."
     },
     {
       "name": "get_assessment",
-      "description": "Poll the authoritative assessment of one assessment.\n\nReturns the highest assessment_version, with the account's current risk. Requires the assess or history scope. While status is pending, or partial with a source still running, the response carries Retry-After; poll no faster than it says. Another workspace's assessment, an unknown or expired assessment and a malformed assessment_id all answer 404 not_found. response_mode=detailed needs the detailed scope."
+      "description": "Poll the authoritative assessment of one assessment.\n\nReturns the highest assessment_version, with the account's current risk. Requires the assess or history scope. While status is pending, or partial with a source still running, the response carries Retry-After; poll no faster than it says. Another workspace's assessment, an unknown or expired assessment and a malformed assessment_id all answer 404 not_found. response_mode=detailed needs the detailed scope. A key with the history scope also gets detail.identifiers, the identifiers the assessment was sent, exactly as received."
     },
     {
       "name": "delete_assessment",
@@ -60,15 +60,19 @@
     },
     {
       "name": "query_assessments",
-      "description": "Page through one account's unexpired assessments.\n\nHistory is a POST with account_id in the body so an account identifier stays out of URLs and access logs. Requires the history scope. Read-only and safe to retry. Newest first, each at its latest assessment_version. An account with no assessments answers 200 with empty items. The cursor is opaque and bound to your workspace and the account; a tampered or expired cursor answers 400 invalid_cursor. response_mode=detailed needs the detailed scope."
+      "description": "Page through one account's unexpired assessments.\n\nHistory is a POST with account_id in the body so an account identifier stays out of URLs and access logs. Requires the history scope. Read-only and safe to retry. Newest first, each at its latest assessment_version. An account with no assessments answers 200 with empty items. The cursor is opaque and bound to your workspace and the account; a tampered or expired cursor answers 400 invalid_cursor. response_mode=detailed needs the detailed scope. Each item's identifiers holds the identifiers its assessment was sent, exactly as received."
+    },
+    {
+      "name": "aggregate_assessments",
+      "description": "Count assessments by time, decision and reason\n\nCount this environment's assessments created in [from,to) by hour, day or week, grouped by up to two of decision, flow, recommendation, band, reason, lane, country, asn and list_match. Each assessment counts once, by its first answer; late results do not count it again and appear in late_flips. Optional metrics add a score histogram per flow and late account decision changes. from and to start a bucket and cover at most 366 days (31 for hour buckets) inside the retention period. The filters are search's and describe the first answer: send them to search with revision first to list a cell's assessments. Counts hold no identifier, stay when an assessment or account is deleted, and report reasons by code and impact band. No Idempotency-Key."
     },
     {
       "name": "search_assessments",
-      "description": "Search assessments across this environment\n\nSearch retained assessments created in [from,to), over at most 366 days. Filters use each assessment decision, rather than the current account decision. Results can change between pages. Keep the assessment version to retrieve the exact decision later. This read does not require an idempotency key."
+      "description": "Search assessments across this environment\n\nSearch retained assessments created in [from,to), over at most 366 days. Filters use each assessment decision, rather than the current account decision. Results can change between pages. Keep the assessment version to retrieve the exact decision later. Each item's detail.identifiers holds the identifiers its assessment was sent, exactly as received. This read does not require an idempotency key."
     },
     {
       "name": "get_assessment_version",
-      "description": "Retrieve an exact assessment revision\n\nRetrieve a retained assessment version with its original decision, policy, flow and depth. This response excludes the current account decision. Current retention and response-mode permissions still apply."
+      "description": "Retrieve an exact assessment revision\n\nRetrieve a retained assessment version with its original decision, policy, flow and depth. This response excludes the current account decision. Current retention and response-mode permissions still apply. detail.identifiers holds the identifiers the assessment was sent, exactly as received."
     },
     {
       "name": "bind_provisional_account",
@@ -88,7 +92,7 @@
     },
     {
       "name": "list_account_assessments",
-      "description": "Page through one account's unexpired assessments.\n\nThe account's history: the same page as POST /v1/assessments/query for this account_id, newest first, each at its latest assessment_version. Requires the history scope. Read-only and safe to retry. An account with no assessments answers 200 with empty items. The cursor is opaque and bound to your workspace and the account; a tampered or expired cursor answers 400 invalid_cursor. response_mode=detailed needs the detailed scope. The account_id travels in the URL; use POST /v1/assessments/query to keep it out of access logs."
+      "description": "Page through one account's unexpired assessments.\n\nThe account's history: the same page as POST /v1/assessments/query for this account_id, newest first, each at its latest assessment_version. Requires the history scope. Read-only and safe to retry. An account with no assessments answers 200 with empty items. The cursor is opaque and bound to your workspace and the account; a tampered or expired cursor answers 400 invalid_cursor. response_mode=detailed needs the detailed scope. Each item's identifiers holds the identifiers its assessment was sent, exactly as received. The account_id travels in the URL; use POST /v1/assessments/query to keep it out of access logs."
     },
     {
       "name": "record_enforcement",
@@ -120,7 +124,7 @@
     },
     {
       "name": "export_account",
-      "description": "Export an account\n\nRetrieve one consistent snapshot of retained customer-accessible records, including assessment revisions, account decisions, feedback, key associations, enforcements and request diagnostics. Secrets, restricted source information, other customers' data and legacy free-text notes are excluded. The same standard/detailed permissions apply. An export exceeding 10000 records or 8 MiB returns export_too_large without a partial result. Historical decisions that were not captured are identified as an exclusion."
+      "description": "Export an account\n\nRetrieve one consistent snapshot of retained customer-accessible records, including assessment revisions, account decisions, feedback, key associations, enforcements and request diagnostics. Each assessment revision's detail.identifiers holds the identifiers that assessment was sent, exactly as received. Secrets, restricted source information, other customers' data and legacy free-text notes are excluded. The same standard/detailed permissions apply. An export exceeding 10000 records or 8 MiB returns export_too_large without a partial result. Historical decisions that were not captured are identified as an exclusion."
     },
     {
       "name": "list_keys",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@noresponse/mcp",
-  "version": "0.6.12",
+  "version": "0.6.14",
   "mcpName": "org.noresponse/mcp",
   "author": "NoResponse",
   "type": "module",