committer

A terminal UI for writing git commit messages with AI. Reviews your staged diff, generates a conventional commit message, lets you edit it, and commits — all from the keyboard (some buttons also support mouse).

Note from the creator: I know editors like VS Code already have this kind of feature built in. I just wanted to build something with OpenTUI — it seemed like a great fit for this.

Install

Option 1: installer (prebuilt binary)

Downloads the latest release binary for your platform (macOS, Linux, or Windows via Git Bash/MSYS2/Cygwin) to ~/.local/bin:

curl -fsSL https://raw.githubusercontent.com/juji/committer-tui/main/install.sh | bash

Supports x64 and arm64. Without a bash shell on Windows, download the matching .exe from the latest release directly.

Option 2: npm

npm install -g committer-tui

Installs a small wrapper that downloads the matching prebuilt binary for your platform on install (same binaries as Option 1).

Option 3: build from source

Requires Bun.

git clone [email protected]:juji/committer-tui.git
cd committer-tui
bun install
bun build --compile src/index.tsx --outfile committer-bin
ln -s "$(pwd)/committer-bin" /usr/local/bin/committer

(bin/committer.mjs is the npm-installed wrapper and expects a downloaded prebuilt binary next to it — not used when building from source.)

Usage

From inside a git repo with staged changes:

committer

Or during development, without linking:

bun dev

Shortcuts

| Key | Action | | -------- | ------------- | | ctrl+g | Open config | | ctrl+y | View history | | ctrl+c | Exit |

Configuration

Press ctrl+g inside the app to add a model, or edit the config file directly at:

• Linux/macOS: ~/.config/committer/config.json
• Windows: %APPDATA%/committer/config.json

Each model needs a provider, model name, and API key:

{
  "models": [
    { "name": "My Model", "provider": "gemini", "model": "gemini-2.0-flash", "apiKey": "..." }
  ]
}

Supported providers: openai, anthropic, gemini, groq, cerebras, mistral, deepseek, together, fireworks, xai, perplexity, requesty, openrouter, ollama, ollama-cloud. ollama runs against a local server (defaults to http://localhost:11434) and doesn't need an API key. ollama-cloud runs against https://ollama.com and needs an API key from your Ollama account.

You can also override the instructions sent to the model with instructionPrefix / instructionSuffix, either from the config screen or directly in the JSON file.

Privacy

API keys are stored only in your local config.json and sent only to the provider you configured for that model, directly from your machine — never to any server we run. Your diff is sent only to the model provider you pick, solely to generate the commit message; nothing is logged or transmitted anywhere else.

Releasing

Pushing a v* tag triggers CI to cross-compile binaries for macOS, Linux, and Windows and attach them to a GitHub Release:

git tag v0.1.0
git push --tags

For a pre-release (beta, rc, etc.), use a tag with a hyphen suffix — CI marks it as a GitHub pre-release automatically, so install.sh (which downloads from releases/latest) skips it until you promote a stable tag:

git tag v0.2.0-beta.1
git push --tags

Development

bun dev          # run with file watching
bun dev:rmconf   # wipe local config