List (and optionally search) the project's memories.
Errors return a
{ message, code }body — see the error reference for thecode→ SDK exception map.
/v1/memoriesLists the project's currently-valid memories (soft-deleted and superseded rows are excluded), newest first. Optional filters narrow by kind, run, trust tier, and review status. When q is set the endpoint runs a lexical full-text search over memory content and returns rank-ordered results; it uses ONLY the lexical index, so it needs no embedding model and works on a deployment that has never pinned one. Browse mode (no q) is keyset-paginated via cursor; search mode returns the first limit results. All reads are scoped to the API key's project.
Authorization
bearerAuth API key sent as a bearer token: Authorization: Bearer <key>.
In: header
Query Parameters
Maximum memories to return (capped at 200).
1 <= value <= 20050Opaque keyset cursor from a previous response's next_cursor, for browse mode (ignored when q is set).
Lexical full-text query over memory content. When set, results are rank-ordered and next_cursor is not returned (search returns the first limit matches).
Restrict to one memory kind.
Value in
- "working"
- "semantic"
- "episodic"
- "procedural"
Restrict to memories distilled from events in this run.
uuidRestrict to one provenance trust tier (e.g. normal).
Restrict to one governance review status (e.g. auto_approved).
Response Body
application/json
application/json
application/json
curl -X GET "https://example.com/v1/memories"{ "memories": [ { "id": "497f6eca-6276-4993-bfeb-53cbbbba6f08", "kind": "string", "content": "string", "created_by_agent": "string", "created_at": "2019-08-24T14:15:22Z", "version": 1, "trust_tier": "string", "review_status": "string", "scope_keys": [ "string" ], "source_event_id": "d8de7367-3c46-4a1a-a2fb-045d0836e007", "run_id": "dded282c-8ebd-44cf-8ba5-9a234973d1ec" } ], "has_more": true, "next_cursor": "string"}{ "message": "string", "code": "string"}{ "message": "string", "code": "string"}