qmd

Index local markdown notes and transcripts for hybrid semantic search.

Updated May 4, 2026

Search Tags:#markdown #rag #search #mcp #knowledge-base #local-ai

Ready to run

npx skills add https://github.com/InverterNetwork/hermes-agent --skill qmd-inverternetwork

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: qmd
Source: https://github.com/InverterNetwork/hermes-agent/tree/main/optional-skills/research/qmd
Command: npx skills add https://github.com/InverterNetwork/hermes-agent --skill qmd-inverternetwork

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires node.js, sqlite.

What problem does it solve?

This skill solves the challenge of fragmented personal knowledge by providing a unified, local search engine that indexes your notes, transcripts, and documentation without relying on cloud services.

Core Features & Use Cases

Hybrid Search: Combines BM25 keyword matching with semantic vector search and LLM-powered reranking for high-precision results.
Local Privacy: Operates entirely on-device, ensuring your sensitive documents and notes never leave your machine.
Use Case: Quickly find specific decisions made in past meetings or locate technical documentation across thousands of markdown files by asking natural language questions.

Quick Start

Ask the agent to search your notes for a specific concept or document by providing a natural language query.

Frequently Asked Questions about qmd

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

FAQPage Schema

How do I search across local markdown notes and meeting transcripts without sending data to the cloud?▼

You can perform local semantic search across markdown notes and meeting transcripts by installing a CLI tool that indexes files entirely on-device. This approach combines BM25 keyword matching with vector embeddings to retrieve information while ensuring your data never leaves your machine.

What is hybrid semantic search and how does it improve knowledge base retrieval?▼

Hybrid semantic search improves knowledge base retrieval by combining traditional BM25 keyword matching, vector embeddings, and LLM-based reranking. This multi-layered approach ensures high-precision results when finding specific decisions or technical documentation across thousands of files.

Do I need Node.js and SQLite to run a local AI search index for markdown files?▼

Yes, Node.js and SQLite are required dependencies to run this local AI search index for markdown files. These components support the CLI tool needed to index your notes and execute hybrid semantic search queries on your device.

Can I use Model Context Protocol for persistent agent access to my personal knowledge base?▼

Yes, you can use Model Context Protocol (MCP) for persistent agent access to your personal knowledge base. MCP integration allows AI agents to continuously query your locally indexed markdown notes and documentation through natural language requests.

What's the best way to find specific decisions from past meeting transcripts using natural language?▼

The best way to find specific decisions in past meeting transcripts using natural language is to use a local search engine with LLM-powered reranking. It processes your natural language query, matches it against indexed transcripts via vector embeddings, and surfaces the most relevant decisions.

Why does my local markdown search rely on BM25 keyword matching instead of just vector search?▼

Relying on BM25 keyword matching alongside vector search ensures high-precision retrieval for your local markdown files. BM25 catches exact term matches that vector search might miss, while LLM-based reranking further refines the final results.