@lerianstudio/matcher-mcp 5.0.0-beta.96 → 5.0.0-beta.97

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.
@@ -14355,7 +14355,7 @@ paths:
14355
14355
  - Ingestion
14356
14356
  /v1/imports/contexts/{contextId}/sources/{sourceId}/extract-document:
14357
14357
  post:
14358
- description: "Stores the uploaded source document tenant-scoped, runs deterministic + AI extraction, and enqueues the resulting transaction CANDIDATES in a human-review queue. AI output is NEVER authoritative: no transaction is reconciled until a human approves the review. The lane is gated by a global kill-switch AND a per-tenant opt-in (fail-closed: a non-opted-in tenant gets 403). The tenant is resolved from the JWT and the context/source from the path — never from the body."
14358
+ description: "Stores the uploaded source document tenant-scoped, runs deterministic + AI extraction, and enqueues the resulting transaction CANDIDATES in a human-review queue. AI output is NEVER authoritative: no transaction is reconciled until a human approves the review. The lane is gated by a global kill-switch AND one process-wide operator consent switch — NOT a per-customer opt-in: granting consent grants AI egress for EVERY customer at once, and while it is off every caller gets 403 (fail-closed). The tenant is resolved from the JWT and the context/source from the path — never from the body."
14359
14359
  operationId: extractDocument
14360
14360
  parameters:
14361
14361
  - description: Context ID
@@ -15255,7 +15255,7 @@ paths:
15255
15255
  tags:
15256
15256
  - Matching
15257
15257
  post:
15258
- description: "Builds AGGREGATE, privacy-safe history features for the context, asks the AI advisor for config-only candidate rules, and enqueues each surviving candidate in a human-review queue. AI output is NEVER authoritative: producing a suggestion CREATES NO rule. A rule is created only when a human APPROVES a suggestion. The lane is gated by a global advisor kill-switch AND a per-tenant opt-in (fail-closed: a non-opted-in tenant gets 403). Tenant from the JWT, context from the path. The egress payload is aggregates only — no raw transaction, money, or PII."
15258
+ description: "Builds AGGREGATE, privacy-safe history features for the context, asks the AI advisor for config-only candidate rules, and enqueues each surviving candidate in a human-review queue. AI output is NEVER authoritative: producing a suggestion CREATES NO rule. A rule is created only when a human APPROVES a suggestion. The lane is gated by a global advisor kill-switch AND one process-wide operator consent switch — NOT a per-customer opt-in: granting consent grants AI egress for EVERY customer at once. Until consent is granted the endpoint fails closed with 403. Tenant from the JWT, context from the path. The egress payload is aggregates only — no raw transaction, money, or PII."
15259
15259
  operationId: suggestMatchRules
15260
15260
  parameters:
15261
15261
  - description: Context ID
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@lerianstudio/matcher-mcp",
3
- "version": "5.0.0-beta.96",
3
+ "version": "5.0.0-beta.97",
4
4
  "type": "module",
5
5
  "description": "Model Context Protocol server for the Matcher reconciliation engine",
6
6
  "license": "Apache-2.0",