Retrieves context from across all memory sources when user references are vague or unclear.
Recall is a specialized context-retrieval skill designed to bridge the gap between vague human references and the AI's memory. It implements a sophisticated fan-out search strategy, scanning daily logs, session transcripts, vault notes, and system logs in parallel to resolve ambiguous terms like 'that thing we discussed' or 'from earlier today.' By prioritizing speed with mandatory timeouts and breadth across diverse data sources, it ensures the agent maintains a high degree of accuracy and continuity without guessing or hallucinating past interactions.
Key Features
01Mandatory 5-second timeouts to ensure high-speed performance
02Keyword extraction for focused cross-repository retrieval
03Automatic trigger detection for ambiguous context phrases
04Multi-source fan-out search across logs, notes, and session transcripts
05Support for searching processed media and Redis state
0649 GitHub stars
Use Cases
01Identifying a specific topic mentioned in a previous session or daily log
02Resolving ambiguous pronouns like 'those photos' or 'that idea'
03Connecting current tasks to historical architecture decisions or vault notes
