qmd

Search local text files with hybrid BM25, vector, and LLM reranking.

Updated Aug 23, 2026
One-click install
npx skills add https://github.com/kotakbiasa/hermes-agent --skill qmd-kotakbiasa
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: qmd
Source: https://github.com/kotakbiasa/hermes-agent/tree/main/optional-skills/research/qmd
Command: npx skills add https://github.com/kotakbiasa/hermes-agent --skill qmd-kotakbiasa

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

BM25 keyword search, semantic vector search, and LLM reranking are combined to enable fast, private local search across notes, transcripts, and documents.

Core Features & Use Cases

  • Hybrid retrieval: keyword + vector + LLM reranking for high-quality results.
  • Local-first: runs entirely on-device without cloud dependencies.
  • MCP integration: exposes tools for Hermes Agent workflow automation and external MCP clients.

Quick Start

Add your collections and run qmd embed to build the index.

Frequently Asked Questions about qmd

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How does hybrid retrieval combine BM25 and vector search for local documents?

Hybrid retrieval merges BM25 keyword matching with semantic vector search and LLM-powered reranking to deliver high-quality local search results across markdown notes and transcripts without cloud dependencies.

Can I run local search across personal markdown notes without sending data to the cloud?

Yes, local search runs entirely on-device without cloud dependencies, ensuring private queries across personal knowledge bases by processing text-based files stored locally on your machine.

How do I build a search index for meeting transcripts using local search?

To build a search index, add your collections of meeting transcripts and documentation, then run the embed CLI tool to generate the vector and BM25 indices required for hybrid retrieval.

Do I need Node.js to perform hybrid retrieval and LLM reranking on local files?

Yes, Node.js is required to execute the CLI tools for collection management, context tagging, and MCP-enabled workflows that facilitate hybrid retrieval and LLM reranking on local files.

Does this local search engine support MCP integration for agent workflows?

Yes, it supports MCP integration by exposing tools for Hermes Agent workflow automation and external MCP clients, enabling automated hybrid retrieval within agent-driven pipelines.

What are the limitations of using LLM reranking for local knowledge base search?

LLM reranking for local knowledge base search requires sufficient on-device compute resources, and the hybrid retrieval engine is limited to querying text-based files like markdown notes and documentation.