Deep Search & Semantic Memory
Integrates local semantic search to help Claude navigate oversized codebases and extensive documentation folders.
Deep Search acts as a long-term memory bridge for your development environment, specifically designed for projects that have outgrown the standard AI context window. By leveraging the tobi/qmd search engine, it enables BM25 and Vector-based retrieval for local files, allowing Claude to find historical specifications, PRDs, and specific code logic across thousands of files. It functions as a non-intrusive power-up that detects when a project is too large to read linearly and suggests indexing for faster, more accurate context retrieval.
Key Features
01Seamless bridge between working memory and long-term document retrieval
02Automated indexing synchronization during the development lifecycle
0343 GitHub stars
04Local semantic search integration using vector embeddings and BM25
05MCP server support for direct integration with AI editors like Cursor and Claude
06Automatic detection of large-scale projects based on file count thresholds
Use Cases
01Searching through vast documentation folders for historical architectural decisions
02Navigating legacy codebases with over 200 source files to find specific logic
03Providing relevant context from old PRDs and meeting notes during new feature planning