qmd

You can search markdown documentation locally by indexing content across repositories and querying it with BM25 and vector search. This returns relevant documents from your local notes and knowledge base without relying on external services.

Lexical search uses BM25 keyword matching to find exact terms in markdown files, while vector search applies semantic matching to retrieve conceptually relevant documents. You can choose lex, vec, or hyde query types to control retrieval behavior.

Yes, you can integrate an MCP server to connect AI assistants with your indexed markdown knowledge base. This allows the assistant to query local notes, policy docs, and API documentation, receiving fast and relevant search results.

The best way to find policy docs and project briefs is indexing markdown content with combined BM25 and vector search. This replaces slow keyword hunts with structured queries that return relevant documents across your local repositories.

The hyde query type enhances vector search by generating hypothetical answers to expand your query semantically. This improves retrieval of relevant markdown documents when searching complex or indirect questions in your knowledge base.

Local search is limited to markdown content and requires proper indexing before querying. Safe defaults and strong error handling manage collection controls, but complex queries across unindexed repositories will not return relevant documents.