qmd

Lightweight SOTA local search engine for AI agents in Rust — BM25 full-text search, vector semantic search, hybrid search with query expansion and reranking, plus an MCP server for AI tool integration.

Crates

Crate		Description
qmd		Core library — indexing, BM25, vector search, hybrid search, embeddings
qmd-cli		CLI tool — collection management, search, RAG question answering
qmd-mcp		MCP server — expose qmd as an AI agent tool

Quick Start

Install the CLI

Shell (macOS / Linux):

curl -fsSL https://sh.qntx.fun/qmd | sh

PowerShell (Windows):

irm https://sh.qntx.fun/qmd/ps | iex

Or via Cargo:

cargo install qmd-cli

CLI Usage

# Add a collection of markdown files
qmd collection add ./docs --name my-docs --mask "**/*.md"

# List collections
qmd ls

# BM25 full-text search
qmd search "query expansion" -n 5

# Vector semantic search (requires embedding model)
qmd models pull          # download default models
qmd embed                # generate embeddings
qmd vsearch "how does reranking work" -n 5

# Hybrid search (BM25 + vector + query expansion + reranking)
qmd qsearch "local search engine for AI"

# Ask a question (RAG)
qmd ask "What search algorithms does qmd support?"

# Get a specific document
qmd get qmd://my-docs/README.md

# Re-index all collections
qmd update

License

Licensed under either of:

• Apache License, Version 2.0 (LICENSE-APACHE or https://www.apache.org/licenses/LICENSE-2.0)
• MIT License (LICENSE-MIT or https://opensource.org/licenses/MIT)

at your option.

Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in this project shall be dual-licensed as above, without any additional terms or conditions.