SmartMemory
GET
/memory/ingest/async/{run_id}

Poll the status of an async ingest run.

status distinguishes abandoned (the content was accepted and is not known to have been stored) from unknown_run_id (no such run). Before the run ledger these were the same reply, so a caller could never learn that content was lost.

Goes through the scoped facade, which refuses a run belonging to another workspace. The previous implementation constructed the transport here and read the status hash directly, so any authenticated caller could read any tenant's run.

Path Parameters

run_id*Run Id

Response Body

application/json

application/json

curl -X GET "https://example.com/memory/ingest/async/string"
null
{
  "detail": [
    {
      "loc": [
        "string"
      ],
      "msg": "string",
      "type": "string",
      "input": null,
      "ctx": {}
    }
  ]
}

Get All Connections

Get all connections (backlinks + forward links). Returns complete bidirectional link graph for this note.

Get Audit Summary

ONTO-RECONCILE-1 Phase 6 — bounded audit aggregation over a time window. Two substrate queries (counts grouped by action+actor; recent-actions LIMIT N). `recent_actions` rows expose only metadata-shaped fields per the privacy contract — caller-supplied content (`evidence_ref`, `before_json`, `after_json`) is excluded. Authentication: routes inherit `ScopePolicy.team_required` from `create_router(...)`'s default — verified at base_router.py:62. Audit data is workspace-scoped via FalkorDB graph_name isolation. Note on `group_by`: today the substrate always groups by `(action, actor)` jointly and folds into `by_action` and `by_actor`. The parameter is accepted (and validated against {"action", "actor"}) for forward compatibility, but additional grouping is not yet implemented.