Search local markdown notes and transcripts with hybrid BM25 and vector retrieval.
npx skills add https://github.com/big4council-prog/b4c-agent --skill qmd-big4council-prog
Please help me install this Agent Skill. Skill: qmd Source: https://github.com/big4council-prog/b4c-agent/tree/main/optional-skills/research/qmd Command: npx skills add https://github.com/big4council-prog/b4c-agent --skill qmd-big4council-prog
It eliminates the pain of searching scattered personal documents by providing fast local retrieval across markdown notes, meeting transcripts, and text-based knowledge bases, with high-quality results through hybrid search and reranking.
Use Case: You remember an action item from a meeting but not the exact wording—use qmd to find the relevant transcript lines and pull the exact document content locally.
Configure a qmd collection for your documents, embed the index, then ask Hermes to run a hybrid query against that collection for the topic you are trying to find.
High-intent search queries and answers about installing and using this skill.
Hybrid local search blends BM25 keyword matching with vector semantic retrieval to locate exact terms and related concepts within markdown notes and transcripts, then applies LLM reranking to surface the most useful passages first.
Searching meeting transcripts locally requires configuring a qmd collection, embedding the index, and running a hybrid query that combines keyword matching and semantic search to retrieve relevant transcript lines and exact document content.
You can expose local knowledge base search tools to an agent via MCP integration by starting the qmd MCP server, which allows agent-driven workflows to query your indexed markdown notes and documentation collections directly.
Yes, local RAG supports vector search and LLM reranking for personal documentation entirely on your machine. A local reranker model evaluates top candidates to prioritize the most relevant passages without requiring external cloud APIs.
Yes, you need a local qmd installation with indexed collections and generated embeddings. Use qmd commands for collection management and embedding generation before querying your markdown notes or starting the MCP server.
A hybrid retrieval pipeline is used because standalone BM25 misses semantic context and standalone vector search can overlook exact terms. Combining both ensures precise phrase matching and concept discovery across local knowledge bases.
afel6Search local markdown and meeting transcripts with hybrid BM25, vector, and reranking retrieval.
JamesFincherCombine BM25 keyword matching, vector search, and LLM reranking for hybrid local retrieval.
steipeteIndex local files and search them with BM25, vector, and hybrid queries.
NITISH-gitbitCombine BM25 keyword matching, vector semantic search, and LLM reranking for local Markdown file retrieval.
