recall

Recall retrieves project context by querying the memories table for project-specific entries and synthesizing them into a concise summary. It automatically filters out deprecated or low-confidence items to deliver reliable historical context for your session.

Memory synthesis for project context works by querying non-deprecated memories from a central database and using a memory-retriever agent to synthesize results. It presents a readable markdown summary to inform your current session context.

You need to query past memories at the start of a session or on-demand when historical context is required. Retrieving synthesized memories provides timely contextual awareness for the current project from a central database.

To exclude deprecated or low-confidence entries when retrieving project memories, the system filters the memories table during the query phase. It specifically queries non-deprecated memories and uses a memory-retriever agent to synthesize only high-confidence results.

If no memories exist when querying project context, the system presents a clear fallback message. The memory-retrieval process checks the database for non-deprecated entries and outputs markdown with a fallback notification when the query returns empty.

Yes, you can use memory retrieval to inform current project context on-demand. The system queries the central database for project-specific memories, synthesizes the results into a concise summary, and presents the context whenever historical awareness is needed.