QMD is a local search engine designed for effortless information retrieval across various document types. It indexes Markdown notes, meeting transcripts, and other knowledge resources, enabling keyword and natural language searches. QMD combines BM25 full-text search, vector semantic search, and LLM-based re-ranking, all executed locally using technologies like node-llama-cpp with GGUF models. By offering a comprehensive and self-contained solution, QMD empowers users to build more capable and contextually aware AI agents.
QMD distinguishes itself through its seamless integration of multiple search techniques — BM25 for speed, vector search for semantic understanding, and LLM reranking for enhanced accuracy. Its focus on local execution ensures privacy and independence from external services. The inclusion of a Model Context Protocol (MCP) server enables tight integration with LLMs like Claude, facilitating powerful agentic workflows. Furthermore, a flexible SDK allows easy embedding into custom Node.js or Bun applications, promoting extensibility.
- Hybrid Search: ** Combines BM25, vector search, and LLM reranking for improved search results.
- MCP Server: ** Offers a standardized interface for integration with various LLMs like Claude.
- SDK & Library: ** Provides comprehensive APIs for Node.js and Bun applications.
- Flexible Configuration: ** Supports configuration via CLI, YAML files, or in-memory settings.
- Contextual Search: ** Allows adding context to documents for improved retrieval accuracy.
- Multiple Search Methods: ** Supports keyword search, semantic search, and pre-expanded queries.
- Document Management: ** Features efficient indexing and retrieval of documents and collections.
QMD is an actively developed project with a growing community. Recent commits and ongoing feature development indicate continuous maintenance and improvement. Comprehensive documentation, including guides and API references, is available. The project's stability is supported by regular updates and a well-defined architecture. The presence of a changelog and active issue tracking further demonstrates its reliability and ongoing evolution.
QMD benefits individuals and teams who need a fast, private, and customizable search solution for their documentation and knowledge base. It streamlines information access, enhances agentic workflows, and offers a valuable alternative to cloud-based search services. By providing powerful search capabilities locally, QMD empowers users to build more capable and self-contained AI-driven applications.