qmd

You can search markdown notes and transcripts locally by indexing your files with a hybrid retrieval engine that runs on-device without cloud dependencies. This enables fast queries across your personal knowledge base.

Hybrid search combines BM25 keyword matching, vector search, and LLM reranking to retrieve documents. This approach improves knowledge base querying by delivering high-quality results for both exact text matches and deeper semantic queries.

Yes, you can run local semantic search across your documentation entirely on-device. The search engine operates locally without cloud dependencies, ensuring your text-based files and personal knowledge base remain private.

Yes, running local text search requires Node.js version 22 or higher, alongside SQLite with extension support. You must also install the search tool to properly index and query your markdown notes and documents.

You index and query files by installing the local search tool, which initializes a SQLite database to support both BM25 and vector search. This setup allows you to perform quick keyword lookups and semantic queries across your collections.

Yes, the local search engine supports per-collection contexts for meeting transcripts and other documents. This allows you to target queries toward specific collections, ensuring context-aware search results across your knowledge base.