Reasoning trail

Report what your agent did, and read it back.

📄️Report execution events

Tell Bairav what your agent did, for applications whose steps do not all go through the chat endpoint: tool calls, database actions, policy checks and so on. Events with the same `trace_id` form one trace you can read back.

📄️Report database queries

Report the SQL your application ran during a chat so Bairav can show real table access on the agent graph. At most 200 queries per request.

📄️Read a trace

Returns every event of one trace in chronological order, scoped to the tenant that owns the gateway key. At most 500 events. The row shape matches the Portal API's `GET /api/lineage/events/traces/{trace_id}`. Use it to render your own reasoning page without a Bairav user session.