The reflect_policy operation configures when and how the reflection pipeline runs automatically. Reflection consolidates raw episodes into insights; the policy determines triggers (e.g., memory count threshold, time interval) and scope (entity, global).
| Protocol | Service | REST Endpoint |
|---|
| gRPC | ReflectService (policy config) | — |
| REST | — | Not yet implemented |
| Parameter | Type | Description |
|---|
trigger_type | enum | memory_count, interval, manual_only. |
memory_threshold | int | Run reflection when entity has N new episodes since last run. |
interval_seconds | int | Run reflection every N seconds (bounded, e.g., max 24h). |
entity_scope | string or null | Entity to scope; null = global. |
max_memories_per_run | int | Cap memories processed per run (bounded). |
max_clusters | int | Cap clusters formed per run. |
- memory_count: After
remember or batch ingest, if entity’s new-episode count >= memory_threshold.
- interval: Background scheduler fires every
interval_seconds.
- manual_only: No automatic triggers; reflection runs only via explicit
reflect call.