reflect:recall

You can retrieve past learnings by querying a knowledge base using the `/reflect:recall` command. This Skill searches a hybrid GraphRAG + BM25 knowledge base to find relevant past project context and prevent repeating previous debugging mistakes.

Knowledge retrieval works through a hybrid GraphRAG and BM25 search mechanism applied to a global learnings knowledge base. It supports local and global search modes to match different context needs, plus a three-step staged retrieval workflow for in-depth querying.

Yes, the Auto-Recall feature automatically retrieves top learnings at the start of a session or project. This ensures relevant past debugging and development context is surfaced immediately without requiring manual query input.

Yes, you need Python dependencies installed to perform the hybrid search. The Skill also requires access to the global learnings knowledge base and the `reflect-kb` dependency to execute the GraphRAG and BM25 retrieval workflows.

Local and global search modes provide different context scopes for information retrieval from the knowledge base. They allow you to tailor the knowledge retrieval process to match either narrow project-specific context or broader cross-project learnings.

You can search for specific debugging context by passing a query string to the retrieval command, such as `/reflect:recall "issue #1234"`. This queries the hybrid knowledge base to surface relevant past learnings related to that specific issue.