Skip to main content
Use an explicit record type to inspect the structured records your agents have saved, including drafts. Reading a draft does not approve it or authorize any action described in it.

List records

The same read is available through erdo_list_knowledge in MCP:
Or the existing REST endpoint:
The response has items, total_count, and an optional next_offset. Each record includes id, organization_id, title, content, type, status, record_type, properties, created_at, and updated_at. type is the original Knowledge object type, such as document or entity; record_type is the exact record classification you requested. properties contains the record’s stored structured properties, including nested objects and arrays.

Read every page

Page size defaults to 20 and is capped at 100. If next_offset is present, make the same request with that offset. Stop only when next_offset is absent. Permissions can produce a short or empty page even when more records follow; do not use items.length or total_count as the stopping condition. total_count retains the underlying Knowledge list’s best-effort count and may include inaccessible records on other pages. For an unchanged result set, records with equal update times and titles have a stable order. Pagination does not create a snapshot: concurrent additions, updates or removals can move records between offset pages.

Status and access

The record filter includes visible non-archived records, including draft and approved, and preserves their status. It does not change approval state, visibility, organization membership, or resource permissions. Use --org or X-Organization-ID to select the organization your credential can access. Do not combine record_type with type, category, or visibility=public (--public in the CLI). Those combinations return an invalid-argument error. If a record’s classification tag disagrees with its typed metadata, the request fails rather than returning a misleading partial page. Without record_type, Knowledge list and search keep their existing behavior: approved memories and skills. See Knowledge for the shared store and API authentication for credentials and organization scope.