Combine BM25, vector search, and LLM reranking for local knowledge search.
npx skills add https://github.com/gqf2008/hermez-ai --skill qmd-gqf2008
Please help me install this Agent Skill. Skill: qmd Source: https://github.com/gqf2008/hermez-ai/tree/main/skills/research/qmd Command: npx skills add https://github.com/gqf2008/hermez-ai --skill qmd-gqf2008
Local, on-device search across notes, docs, and meeting transcripts using a hybrid engine that blends BM25, vector search, and LLM reranking for fast, relevant results.
Query your local notes and transcripts with qmd to retrieve relevant documents quickly and rank them by relevance.
High-intent search queries and answers about installing and using this skill.
To search local notes by concept, you can use a hybrid retriever that combines BM25 keyword search, vector search, and LLM reranking. This blends semantic understanding with exact matching to locate relevant documents across your local knowledge base.
Hybrid retrieval combines BM25, vector search, and LLM reranking to improve local knowledge base search. It merges exact keyword matching with semantic understanding, then uses a language model to rank results for higher relevance.
Yes, you can search personal documents on macOS without cloud dependencies using a local search engine. It operates entirely on-device across local knowledge bases, notes, and meeting transcripts, ensuring data privacy.
Yes, MCP integration works for local document search and retrieval. It connects the local hybrid search engine to agent workflows, allowing tool access to query personal knowledge bases and retrieve ranked results.
The best way to find a specific meeting note is using a local hybrid search engine. It applies BM25, vector search, and LLM reranking to quickly locate project documents or transcripts by concept rather than exact phrasing.
BM25 and vector search for personal notes require local installation on macOS or Linux. While LLM reranking improves result accuracy, the system operates entirely locally without cloud dependencies, which may limit processing speed on lower-end devices.
steipeteIndex local files and search them with BM25, vector, and hybrid queries.
ovachieverQuery ClinicalTrials.gov API v2 and export trial data as JSON or CSV.
tawf-labsBuild retrieval-augmented generation workflows that ground LLM answers in external documents.
AbilityaiRebuild the FAISS vector index for updated semantic search accuracy.
