Search markdown collections using BM25, vector, and hybrid re-ranking.
npx skills add https://github.com/pbdeuchler/llm-plugins --skill howto-qmd
Please help me install this Agent Skill. Skill: howto-qmd Source: https://github.com/pbdeuchler/llm-plugins/tree/main/plugins/tooling/skills/howto-qmd Command: npx skills add https://github.com/pbdeuchler/llm-plugins --skill howto-qmd
qmd helps you search markdown knowledge bases, documentation, and notes using a combined BM25 keyword search, vector semantic search, and LLM-powered hybrid ranking to surface relevant context.
Ask qmd to perform a search across your indexed markdown collection.
High-intent search queries and answers about installing and using this skill.
You can search local markdown documents by configuring qmd collections and running the embed command before executing vsearch or query to surface relevant context using BM25 and vector semantic search.
Building context from scattered documentation is achieved by applying qmd's hybrid re-ranking across local collections of markdown files, combining BM25 keyword search and vector semantic search to answer questions or gather background material.
Markdown search results can be output in multiple formats, allowing you to export retrieved context as JSON, CSV, or standard MD formats directly from your qmd queries.
Yes, you must run the embed command before executing vsearch or query operations on your markdown knowledge base to ensure the vector semantic search and hybrid re-ranking function correctly.
Hybrid re-ranking improves markdown knowledge base search by combining BM25 keyword search with vector semantic search, using an LLM-powered approach to surface more relevant context than basic keyword matching alone.
Limitations of searching markdown knowledge bases with qmd include the requirement to configure collections beforehand and the necessity of running the embed step prior to any vsearch or query execution.
cdeistopenedIndex and search local Markdown collections with BM25 and vector embeddings.
imehrLog AI executions and learn from feedback using JSONL logs and JSON recommendations.
projectedanxIndexes and queries local markdown files and transcripts via hybrid BM25, vector embeddings, and LLM reranking search engine.
tc9011Search personal markdown knowledge bases with hybrid BM25 and vector retrieval.
