Index local files and search them with BM25, vector, and hybrid queries.
npx skills add https://github.com/steipete/clawdis --skill qmd
Please help me install this Agent Skill. Skill: qmd Source: https://github.com/steipete/clawdis/tree/main/skills/qmd Command: npx skills add https://github.com/steipete/clawdis --skill qmd
This Skill helps you quickly index and search local documents, enabling fast retrieval with hybrid BM25 lexical matching and vector-based ranking, with optional reranking.
Use the qmd CLI to index local files: qmd collection add /path --name docs --mask "**/*.md" qmd update qmd search "deployment plan"
High-intent search queries and answers about installing and using this skill.
File indexing enables fast retrieval by building a searchable index of your documents. qmd indexes local files using BM25 lexical matching and vector embeddings, then returns ranked results instantly across thousands of files with a single query command.
BM25 performs keyword-based lexical matching for exact phrase matches; vector search finds semantically similar content. qmd combines both in hybrid search to catch exact matches and conceptually related documents, improving recall and relevance simultaneously.
Yes. qmd supports local repositories of Markdown, code, and text files. You configure file masks (e.g., **/*.md) to target specific file types, then index and query across your entire repository structure stored under ~/.cache/qmd.
Ollama is required for embeddings and reranking. If you configure OLLAMA_URL, qmd uses Ollama to generate vector embeddings; without it, you can still perform BM25 lexical search but lose vector and reranking capabilities.
Use three commands: qmd collection add /path --name docs --mask "**/*.md" to add files, qmd update to build the index, then qmd search "query" to retrieve results. The index stores locally under ~/.cache/qmd by default.
Yes. qmd supports MCP mode via the mcp command, enabling integration with Model Context Protocol workflows for embedding qmd search capabilities into larger automation and AI tool pipelines.
johncarpenterSearch, retrieve, and index local Markdown files using the QMD CLI.
D3HunterFormat code review findings into severity-sorted JSON with rubric metadata.
steipeteIndex local files and search them with BM25, vector, and hybrid queries.
zach-sourceSearch codebases semantically using natural language queries with local embeddings.
