CommitAI

CommitAI is a command-line tool built in Go that leverages Large Language Models (LLMs) to analyze your Git code changes and automatically generate semantic commit messages.

Features

🎯 Core Features

• 🤖 Automatic commit message generation - AI-powered commit analysis
• 🌐 Dual processing modes - Local and remote LLM processing
• 🔄 Easy configuration management - Simple setup and updates
• 🌍 Multi-language support - Commit messages in any language
• 🏷️ Custom commit tags - Force specific commit types

✨ New in Version 2.0

• 📋 Multiple commit formats - Support for native Git or custom standards
• 🎨 Personal prompts - Define your own commit patterns and styles
• 🔌 Universal model support - Any local model or free web version compatibility
• 📦 Easy installation - Available via NPM or compiled binaries
• 🔧 Native Git integration - Enhanced Git support with advanced customization
• 🖥️ Improved interface - Better menu navigation and user experience
• ⚡ Performance boost - Faster execution and optimized CLI performance

Installation

CommitAI offers multiple installation methods to suit your preferences:

🚀 Quick Install (Recommended)

Option 1: NPM Global Install

npm install -g commit-ai-hublast

Option 2: From Repository

git clone https://github.com/HublastX/Commit-IA
cd Commit-IA
npm install -g

🔧 Manual Compilation

Prerequisites:

• Go 1.24 or higher
• Node.js 20+ (for NPM installation)

Steps:

1. Clone the repository:

   git clone https://github.com/HublastX/Commit-IA
   cd Commit-IA

2. Compile for your platform using Go:

   go build -o commitai ./cmd

3. Move binary to your PATH or use local installer scripts

🛠️ Alternative Installation Methods

Legacy Installer Scripts (Linux/macOS):

chmod +x ./install.sh
./install.sh

Legacy Installer Scripts (Windows):

.\install.bat

🔧 Troubleshooting Installation

If you encounter build errors during manual compilation, install the required development packages:

1. Install Scoop package manager (PowerShell):

   Set-ExecutionPolicy RemoteSigned -scope CurrentUser
   iwr -useb get.scoop.sh | iex

2. Install dependencies:

   scoop install mingw

sudo apt update && sudo apt install -y \
    gcc libc6-dev libx11-dev xorg-dev \
    libxtst-dev libpng-dev libxcursor-dev \
    libxrandr-dev libxinerama-dev libdbus-1-dev \
    tesseract-ocr

sudo pacman -Syu && sudo pacman -S --needed \
    gcc glibc libx11 xorg-server-devel \
    libxtst libpng libxcursor libxrandr \
    libxinerama dbus tesseract

Usage

After using git add to stage your changes, you can use commitai to generate commit messages.

Basic Command

commitai

Operating Modes

commitai offers two operating modes:

1. Remote Web Mode - Access LLMs remotely without additional configuration (might be slower)
2. Local Mode - Run the LLM API locally using Docker (requires provider configuration)

To switch between modes or update configuration:

commitai --update

Additional Options

| Option | Description | Example | | ---------- | ------------------------------------ | ----------------------------------------- | | -d | Add additional context | commitai -d "Added login functionality" | | -l | Specify commit language | commitai -l "English" | | -t | Force specific commit tag | commitai -t "feat" | | --update | Update configuration or switch modes | commitai --update |

Examples

Generate a commit with additional context:

commitai -d "Created user login feature with OAuth support"

Generate a commit message in English:

commitai -l "English"

Force a specific commit tag:

commitai -t "feat"

Update configuration or switch between local and web modes:

commitai --update

Local LLM Configuration

When selecting the Local mode, simply configure through the CLI:

1. Run commitai --update to access configuration
2. Select Local Mode when prompted
3. Choose your LLM provider (Google, OpenAI, Anthropic, etc.)
4. Select an available model for your chosen provider
5. Enter your API key for the provider

That's it! No Docker setup required - the tool will handle everything automatically.

Configuration Details

• Web mode uses Google's gemini-flash-2 model by default
• All LLM configurations are stored in the Bot directory
• Custom configurations (provider, model, API token) are managed in Bot/config/config.json

Contributing

Contributions are welcome! Feel free to open issues or submit pull requests.

We recommend using Husky for commit validation:

1. Ensure Node.js 22 is installed
2. Run npm install to set up Husky, which will validate your commits before submission

License

This project is licensed under the Apache-2.0 license.