QMD is a command-line search engine for local markdown documents, meeting notes, and knowledge bases that combines full-text search, vector semantic search, and LLM re-ranking entirely on-device.
The tool addresses the problem of searching across scattered personal documentation by indexing markdown files and supporting multiple search approaches simultaneously. It runs BM25 full-text search, vector semantic search, and LLM re-ranking locally using node-llama-cpp with GGUF models, then combines results via reciprocal rank fusion. The tool outputs results in formats designed for integration with AI agents, including JSON and file listings, and exposes an MCP server for tighter integration with Claude and other AI systems.
QMD suits developers and knowledge workers who maintain local markdown collections and want search without cloud dependencies or external API calls. The MCP server integration makes it particularly valuable for agentic workflows where an AI assistant needs to retrieve and reason over your documentation. The tool is designed for scenarios where you control the data and want semantic search capabilities without sending content elsewhere.
The project receives substantial real-world adoption, as evidenced by most open issues coming from outside users rather than the core team. Responses to issues and pull requests typically arrive within one to two weeks. Development activity concentrates on bug fixes, enhancements, and documentation improvements.
