qmd: An on-device search engine for everything you need to remember.
What problem it solves
Local notes, documentation, knowledge bases, and meeting transcripts need to remain searchable without sending their contents to a remote service. QMD addresses this by combining keyword retrieval, semantic search, and local LLM reranking for documents indexed on the user's device.
Key features
Hybrid local search — Queries can use BM25 full-text search, vector semantic search, or a hybrid pipeline that expands the query, fuses results with Reciprocal Rank Fusion, and applies LLM reranking.
Source: README
Collections and contextual metadata — Users can organize notes, documents, and transcripts into collections and attach hierarchical context that is returned with matching documents.
Source: README
Document and batch retrieval — Documents can be retrieved by path or document ID, while batch retrieval supports glob patterns, comma-separated lists, and document IDs.
Source: README
Agent and library integrations — JSON and file-oriented outputs support agentic workflows, an MCP server exposes search and retrieval tools, and the SDK can embed QMD in Node.js or Bun applications.
Source: README, src
Who it may suit
This may suit developers and knowledge workers who want local keyword and semantic search across Markdown documents, notes, documentation, and meeting transcripts. It may also suit teams building MCP- or SDK-based retrieval into agentic applications.
Topics to discuss
How should hybrid search balance exact matches, semantic retrieval, and LLM reranking?
Which local models and metadata-filtering workflows best support multilingual collections?
What MCP and SDK workflows would make local document retrieval more useful for agents?
Written by ReadmeX AI from the project's README, source and public GitHub data; it may contain mistakes.
Read the original README https://github.com/tobi/qmd#readme
ReadmeX opened this community for a trending project; it is not run by the project. Maintainers can claim it from the community page.
Comments
…