qmd

To search markdown notes and meeting transcripts locally, you can use qmd for on-device retrieval. It combines BM25 keyword matching and vector semantic retrieval to find information across personal document collections without relying on external services.

Yes, running local semantic search requires Node.js version 22 or higher and SQLite with extension support. These dependencies are necessary to execute the qmd command-line tool and its pre-downloaded embedding models on your system.

You can use qmd for knowledge base search through both command-line usage and MCP integration. This allows seamless operation from terminals, chat agents, or automated workflows to locate information across your personal documents.

The best way to retrieve information from large collections of personal documents on macOS is using a hybrid search approach. qmd combines BM25 keyword matching, vector semantic retrieval, and LLM reranking to deliver high-quality search results locally.

Local document search supports both macOS and Linux systems. qmd is designed to apply semantic retrieval and keyword matching to text files and markdown notes across these operating systems without requiring external cloud services.