Retrieve relevant markdown content using hybrid BM25 and vector search.
npx skills add https://github.com/compozy/agh --skill qmd-compozy
Please help me install this Agent Skill. Skill: qmd Source: https://github.com/compozy/agh/tree/main/.agents/skills/qmd Command: npx skills add https://github.com/compozy/agh --skill qmd-compozy
Users need fast answers from scattered markdown notes and documentation without manually hunting through files.
Run the command qmd query "How does the rate limiter handle burst traffic" to get a ranked answer with supporting snippets from your embedded markdown collections.
High-intent search queries and answers about installing and using this skill.
Semantic search over markdown notes combines lexical BM25 keyword matching with vector semantic search and optional hypothetical-answer reranking to retrieve relevant information when exact keywords are unknown.
The best way to search markdown documentation without exact keywords is using hybrid retrieval, which combines vector semantic search with BM25 lexical matching to improve recall across knowledge bases.
You run a query against multiple embedded markdown collections using query-time controls, applying collection filtering and specifying limit and expansion behavior to scope the retrieved results.
Yes, you can use an MCP server or an HTTP API for markdown knowledge base retrieval, allowing agents to query embedded markdown collections and fetch supporting document snippets on demand.
Yes, hybrid search requires a local index created with collection and embedding steps before you can query embedded markdown content for question answering or documentation lookup.
BM25 lexical search alone limits documentation lookup recall when users lack exact keywords, which is why hybrid retrieval adds vector semantic search and hypothetical-answer reranking to improve results.
caiazzopIndex and search local files using BM25, vector embeddings, and reranking.
fcakyonSearch the web and extract content from URLs via Tavily APIs.
C0ntr0lledCha0sConfigure MCP servers for Claude Code plugins via .mcp.json or plugin.json.
iia-argReplace grep-based memory search with hybrid BM25 and vector semantic search.
