Skip to content

Getting Started

Install

Stable Release

bash
cargo install commitbee

Development Version

If you want access to the latest features weeks early (warning: can be unstable, no guarantees):

bash
cargo install --git https://github.com/Sephyi/commitbee --branch development commitbee

Or build it yourself from source:

bash
git clone https://github.com/sephyi/commitbee.git
cd commitbee
cargo build --release
# Binary at ./target/release/commitbee

Requirements

  • Rust 1.94+ (edition 2024)
  • An LLM Provider:
    • Local (Default): Ollama running locally with a model pulled (e.g., ollama pull qwen3.5:4b).
    • Cloud: An API key for Anthropic or OpenAI (see LLM Providers).

More providers will be added natively in the future. For now, you can hook into almost any other LLM (like Gemini, Mistral, or custom
endpoints) by using the openai provider and setting the openai_base_url in your config to point to an OpenAI-compatible proxy.

First Run

The most important concept in CommitBee is that you must always stage your files first. CommitBee only looks at your staging area,
giving you complete control over which files are included or ignored in the generated commit message.

bash
# Explicitly stage the files you want
git add src/my_change.rs

# Generate a commit message from staged context
commitbee

That’s it. Zero configuration needed if Ollama is running with qwen3.5:4b.

CommitBee will analyze your staged changes, extract semantic information via tree-sitter, send a structured prompt to the LLM, validate
the output, and present you with a commit message to approve.

Quick Config

Want to customize things? Create a config file:

bash
commitbee init

This creates a config at your platform’s standard location (run commitbee doctor to see where). Edit it to change the model, provider, or formatting preferences.

🔥 Next Step: Ready to level up? Check out the Recommended Setup guide for the official configuration we use for the highest quality commit messages.