pycommit-ai 0.6.1

AI-generated Git commits

pycommit-ai

AI-powered Git commit message generator — get meaningful, conventional commit messages from multiple AI providers simultaneously.

Inspired by aicommit2, rewritten in Python.

✨ Features

• 🤖 Multi-provider support — Gemini, OpenAI, Groq, and OpenRouter
• 📝 Conventional Commits & Gitmoji — choose your preferred commit style
• 🌍 Multilingual — generate messages in any locale (e.g., en, pt, es)
• ⚡ Concurrent generation — queries all configured providers at once
• 🚀 Pull Request Generation — generates PR descriptions from branch diffs
• 🔧 Flexible configuration — INI file, environment variables, or CLI flags
• ⌨️ Quick alias — use pyc as a fast shortcut

📦 Installation

Requirements: Python ≥ 3.13, uv

Global Install

uv tool install git+https://github.com/dnilsonc/pycommit-ai.git

(The pycommit-ai command and pyc short alias will be available globally)

Update/Uninstall:

uv tool upgrade pycommit-ai
uv tool uninstall pycommit-ai

Alternative methods

Method	Command
pipx	pipx install git+https://github.com/dnilsonc/pycommit-ai.git
pip	pip install git+https://github.com/dnilsonc/pycommit-ai.git
Local dev	git clone ... && cd pycommit-ai && uv sync

🚀 Quick Start

1. Configure an AI provider (or multiple)

# Gemini (default model: gemini-2.5-flash)
pycommit-ai config set GEMINI.key=YOUR_GEMINI_API_KEY

# OpenAI (default model: gpt-4o-mini)
pycommit-ai config set OPENAI.key=YOUR_OPENAI_API_KEY

# Groq (default model: llama3-8b-8192)
pycommit-ai config set GROQ.key=YOUR_GROQ_API_KEY

# OpenRouter (default model: google/gemini-2.0-flash-001)
pycommit-ai config set OPENROUTER.key=YOUR_OPENROUTER_API_KEY

2. Stage your changes and run

git add .
pycommit-ai

3. Generate a Pull Request description

pyc --pr

(Optionally, set a custom PR template: pyc config set PR.templatePath=~/my-pr-template.md)

🏳️ CLI Flags

Flag	Short	Description
--locale	-l	Locale for the commit message (e.g., pt, en, es)
--generate	-g	Number of messages to generate per model
--all	-a	Stage all changed files before generating
--type	-t	Commit message style: conventional or gitmoji
--confirm	-y	Auto-commit with the first generated message (no prompt)
--dry-run	-d	Show generated messages without committing
--copy	-c	Copy the selected message to clipboard instead of committing
--exclude	-x	Files to exclude from the diff (can be used multiple times)
--print-prompt	-p	Don't use AI, just print and copy the generated prompt

Examples

pycommit-ai -g 3 -l pt             # 3 messages per model in Portuguese
pycommit-ai -a -t gitmoji -y       # Gitmoji, stage all, auto-commit
pycommit-ai -c                     # Copy message to clipboard instead of committing
pyc -x "*-lock.json"               # Exclude lock files from diff
pyc --pr                           # Generate a PR description
pyc --pr --print-prompt            # Print the prompt without calling the AI

Note: Common lock files (uv.lock, package-lock.json, etc.) are excluded by default.

⚙️ Configuration

Configuration is stored in ~/.config/pycommit-ai/config.ini (follows XDG).

Config Commands

pycommit-ai config set KEY=VALUE          # Set a value
pycommit-ai config get KEY                # Get a value
pycommit-ai config list                   # List all keys
pycommit-ai config del KEY                # Delete a key
pycommit-ai config path                   # Show config file path

Available Settings

Setting	Default	Description
locale	en	Language for generated messages
generate	1	Number of messages per model
type	conventional	Commit style (conventional / gitmoji)
maxLength	50	Max character length for the subject line
timeout	10000	Request timeout in milliseconds
maxTokens	2048	Max tokens for AI response
temperature	0.7	AI creativity (0.0 – 1.0)
topP	1.0	Nucleus sampling parameter
systemPrompt	—	Custom system prompt (inline)
systemPromptPath	—	Path to a custom system prompt file
PR.template	—	Custom PR template (inline markdown)
PR.templatePath	—	Recommended: Path to a .md custom PR template file

Provider Settings

Each provider supports key and model (comma-separated for multiple models):

# Use multiple models for a single provider
pycommit-ai config set OPENAI.model=gpt-4o,gpt-4o-mini

# Custom API URL (OpenAI-compatible endpoints)
pycommit-ai config set OPENAI.url=https://your-custom-api.com
pycommit-ai config set OPENAI.path=/v1/chat/completions

Environment Variables

API keys can also be set via environment variables:

export GEMINI_API_KEY=your-key
export OPENAI_API_KEY=your-key
export GROQ_API_KEY=your-key
export OPENROUTER_API_KEY=your-key

Use PYCOMMIT_AI_CONFIG_PATH to override the config file location:

export PYCOMMIT_AI_CONFIG_PATH=/path/to/custom/config.ini

Config Priority

Settings are resolved in this order (highest priority first):

1. CLI flags (--locale, --generate, etc.)
2. Environment variables (GEMINI_API_KEY, etc.)
3. Config file (~/.config/pycommit-ai/config.ini)
4. Defaults

🧪 Development

# Install dev dependencies
uv sync --group dev

# Run tests
uv run pytest -v

# Run the CLI locally
uv run pycommit-ai --help

📄 License

MIT