Семантический поиск по базе знаний сеансов
productivity
recall is a Claude Code skill that runs semantic search over a knowledge base built from past working sessions. It queries PostgreSQL with BGE embeddings by executing `recall_learnings.py` and returns the top 5 matches, each annotated with learning type, confidence level, and session context. Three search modes are available: the default hybrid mode, `--vector-only` for higher precision, and `--text-only` for faster lookups; the result count is controlled with `--k`. Developers use it to surface hook patterns, TypeScript error fixes, or wizard installation notes from previous sessions without manually scanning history.
- #semantic-search
- #memory-retrieval
- #embeddings
- #knowledge-management