Use when retrieving memories from a long-term store to inject into an agent's context: rewrite queries from the current conversation, run hybrid retrieval (keyword + semantic + recency), fuse scores, and inject within a token budget. Triggers on memory retrieval, hybrid retrieval, memory injection, memory retrieval, hybrid search, context injection, token budget, query rewriting, recency weighting, knowledge retrieval, RAG, long-term memory. NOT for building the store or managing its lifecycle — use memory-architect or memory-manager.
