memory-query

To search past conversations and retrieve historical context, query the agent-memory system using simple prompts like "What did we discuss about X?" or commands such as /memory-search <topic> to surface relevant past discussions and decisions.

Explainable retrieval works by applying tiered retrieval with automatic fallbacks and intent-based routing, returning results that include the tier used, selected paths, and grip-based excerpts to justify the historical context surfaced.

Yes, you can run time-bounded queries to recall conversations from specific periods using commands like /memory-recent --days 7 to fetch context-aware historical discussions and sessions within that defined timeframe.

Yes, you need a running memory-daemon service to use conversation search and retrieval, as the system queries stored historical context through grip-based excerpts and TOC navigation supported by this background service.

The best way to navigate large conversation histories is using grip-based excerpts and TOC navigation, which outputs structured bullets with grip identifiers to surface relevant past discussions and reduce context reassembly effort.

Limitations of intent-based routing for memory search include dependency on automatic fallbacks across tiered retrieval, which requires a running memory-daemon service to successfully route time-bounded queries and return explainable results.