Vector Memory

Gives Claude semantic memory for documents by wrapping a Redis vector store. You can save files and retrieve them later using natural language queries instead of exact filenames or paths. The semantic search means you can ask for "that Python script about database migrations" and get relevant results even if you never used those exact words in the filename. Reach for this when you want Claude to remember and recall documents across conversations without manually tracking what you've shared. Requires a Redis instance with vector search support running, so there's some setup involved, but you get persistent memory that survives beyond single chat sessions.