GitGen

GitHub release (latest by date) GitHub license

AI-Powered Git Commit Message Generator

A multi-model AI commit message generator that seamlessly integrates with your Git workflow. Configure multiple AI providers, switch between models instantly, and generate contextual commit messages that capture the essence of your changes.

Key Features

Multi-Model Architecture - Configure unlimited AI models from different providers
Instant Model Switching - Use aliases like @fast, @smart, @free to switch models on the fly
Secure Configuration - API keys encrypted using platform-specific data protection
Cost Tracking - Monitor token usage and costs per model with multi-currency support
OpenAI-Compatible APIs - Works with OpenAI, Anthropic, Azure, Google, Groq, local models, and any provider offering OpenAI-compatible endpoints
Smart Parameter Detection - Automatically adapts to provider-specific API variations
Zero Dependencies - Single executable, no runtime requirements
Interactive Configuration - Guided wizard with testing and validation
Partial Alias Matching - Type partial aliases like @ult to match @ultrathink
Per-Model System Prompts - Customize behavior for each model

Quick Start

# Download from releases and add to PATH
# https://github.com/stewartcelani/GitGen/releases

# First-time setup - launches configuration wizard
gitgen config

# Generate a commit message with default model
gitgen

# Use a specific model via alias
gitgen @smart

# Guide the generation with custom instructions
gitgen "focus on security changes"
gitgen @fast "explain the refactoring"

Command Reference

$ gitgen help
GitGen - AI-Powered Git Commit Message Generator

Usage:
  gitgen                       Generate commit message with default model
  gitgen [prompt]              Generate with custom prompt
  gitgen @<model>              Generate with specific model or alias
  gitgen [prompt] @<model>     Generate with custom prompt and model
  gitgen @<model> [prompt]     Alternative syntax
  gitgen [command] [options]   Run a specific command

Examples:
  gitgen
  gitgen "must be a haiku"
  gitgen @fast                 # Use your fast model
  gitgen @free                 # Use free model for public repos
  gitgen -p @fast              # Preview model selection and cost
  gitgen "focus on security" @ultrathink
  gitgen @sonnet "explain the refactoring"

💡 Tip: Configure a free model as @free to save money on public repositories
   where sending code to free APIs doesn't matter.

Options:
  -d, --debug            Enable debug logging
  -p, --preview          Preview mode - show what would happen without calling LLM
  -v, --version          Show version information
  -?, -h, --help         Show help and usage information

Commands:
  config                 Run the interactive configuration menu
  help                   Display help information

Installation

Download Binary

Download the appropriate release for your platform from Releases
Extract the archive to a directory in your PATH
Run gitgen config to set up your first model

Build from Source

Requirements: .NET 9.0 SDK

git clone https://github.com/stewartcelani/GitGen.git
cd GitGen
./publish.ps1  # Creates platform-specific executables in dist/

Complete User Experience Guide

Main Configuration Menu

Running gitgen config opens the main configuration interface:

╔════════════════════════════════════════╗
║         GitGen Configuration           ║
╚════════════════════════════════════════╝

1. Add new model
2. Manage models (3 configured)
3. Test models
4. App settings
5. Reset all configuration
0. Exit

Select option: _

Adding a New Model - Complete Wizard Flow

The configuration wizard guides you through 10 comprehensive steps:

Step 1: Model Name

🎉 Welcome to the GitGen Multi-Model Configuration Wizard
This will guide you through setting up a new AI model configuration.

Step 1: Choose a name for this model configuration.
This is a friendly name to identify this configuration, NOT the model ID the provider uses.
Examples: 'gpt-4-work', 'sonnet', 'kimik2', 'llama-local'
Enter model name: claude-work

Step 2: Aliases Configuration

Step 2: Configure aliases for quick access (optional).
Aliases allow you to quickly reference models with memorable shortcuts.

Examples:
  @ultrathink - For complex reasoning tasks
  @sonnet    - For general coding tasks
  @free      - For public repos where privacy isn't an issue

💡 Tip: Configure a free model as @free to save money on public repositories
⚠️  Important: Avoid setting a free model as your default

Enter aliases (comma-separated) [@claudework]: @claude, @work, @smart
✅ Configured aliases: @claude, @work, @smart

Step 3: Description

Step 3: Add a description for this model (optional).
This helps you remember what this model is best used for.
Enter description [none]: Company API key - high capability model for complex tasks

Step 4-5: Provider Configuration

Step 4: Select your provider's API compatibility type.
  1. OpenAI Compatible (e.g., OpenAI, Azure, Groq, Ollama)
Enter your choice: [1] 1

Step 5: Select your specific provider preset.
  1. OpenAI (Official Platform)
  2. Custom Provider (API Key required, e.g., Azure, Anthropic, Google, OpenRouter, Groq)
  3. Custom Provider (No API Key required, e.g., Ollama, LM Studio)
Enter your choice: [1] 2
Enter the provider's chat completions URL: https://api.anthropic.com/v1/chat/completions
Provider name [anthropic.com]: Anthropic
Enter the model ID used by the provider's API: claude-sonnet-4-20250514
Enter the provider's API Key: **************************************************

Step 6-7: Configuration & Testing

Step 6: Configure maximum output tokens.
ℹ️ Suggested: 2000 tokens (Standard model - lower limit sufficient)
Enter max output tokens: [2000] 3000

Step 7: Test the configuration.
Testing your configuration and detecting optimal API parameters...
🧪 Testing LLM connection...
🔗 Using Anthropic provider via https://api.anthropic.com/v1/chat/completions

✅ LLM Response:
"Hello! I'm working great. Ready to help you generate meaningful commit messages!"

Generated with 24 input tokens, 18 output tokens (42 total) • 73 characters
🎉 Configuration test successful!

Step 8-10: Optional Configuration

Step 8: Configure pricing information (optional).
Select currency:
  1. USD ($)
  2. EUR (€)
  3. GBP (£)
  4. AUD (A$)
  5. Other
Enter your choice: [1] 1
Input cost per million tokens [0]: 3
Output cost per million tokens [0]: 15

Step 9: Configure custom system prompt (optional).
Example: 'Always use conventional commit format'
Enter custom system prompt: Focus on why changes were made, not just what changed

Step 10: Review configuration summary.
📋 Model Configuration Summary:
   Name: claude-work
   Aliases: @claude, @work, @smart
   Description: Company API key - high capability model for complex tasks
   Type: openai-compatible
   Provider: Anthropic
   URL: https://api.anthropic.com/v1/chat/completions
   Model ID: claude-sonnet-4-20250514
   Max Output Tokens: 3000
   Pricing: Input: $3.00/M tokens, Output: $15.00/M tokens
   System Prompt: Focus on why changes were made, not just what changed

Save this model configuration? [y]: y
✅ Model 'claude-work' saved successfully!

Model Management Interface

The model management submenu provides comprehensive model control:

═══ Model Management ═══

1. List models
2. Set default model
3. Edit model (aliases, tokens, etc.)
4. Delete model
0. Back to main menu

Listing Models

═══ Configured Models ═══

  claude-work ⭐ (default)
    Type: openai-compatible | Provider: Anthropic | Model: claude-sonnet-4-20250514
    URL: https://api.anthropic.com/v1/chat/completions
    Temperature: 0.3 | Max Output Tokens: 3,000
    Note: Company API key - high capability model for complex tasks
    Aliases: @claude, @work, @smart
    Pricing: Input: $3.00/M tokens, Output: $15.00/M tokens
    Last used: 2025-07-28 10:45 AM

  groq-fast
    Type: openai-compatible | Provider: Groq | Model: llama-3.1-70b-versatile
    URL: https://api.groq.com/openai/v1/chat/completions
    Temperature: 0.3 | Max Output Tokens: 2,000
    Note: Ultra-fast inference for quick commits
    Aliases: @fast, @quick, @groq
    Pricing: Input: $0.59/M tokens, Output: $0.79/M tokens
    Last used: 2025-07-28 09:30 AM

Generation Output Examples

When generating commit messages, GitGen provides rich feedback:

$ gitgen
Found 5 changed files
🔗 Using claude-work (claude-sonnet-4-20250514 via Anthropic)
✅ Generated Commit Message:
"Refactor authentication middleware to support JWT refresh tokens, add concurrent device handling, and improve error responses with detailed status codes for better debugging"

Generated with 3,847 input tokens, 38 output tokens (3,885 total) • 178 characters
Estimated cost: $0.07 USD

📋 Commit message copied to clipboard.

With cost preview for large diffs:

$ gitgen @ultrathink
Found 47 changed files
⚠️  Large diff detected: ~18,000 tokens

💰 Estimated cost:
   • Input: ~$0.18
   • Output: ~$0.02
   • Total: ~$0.20

Continue? (y/N): y

Provider Configuration Examples

OpenAI Configuration

Step 5: Select your specific provider preset.
Enter your choice: [1] 1
Enter the model ID: [gpt-4o-mini] gpt-4-turbo
Enter your OpenAI API Key: sk-**************************************************

Claude (Anthropic) Configuration

Step 5: Select your specific provider preset.
Enter your choi

GitGen

Install / Use

README

GitGen

AI-Powered Git Commit Message Generator

Key Features

Quick Start

Command Reference

Installation

Download Binary

Build from Source

Complete User Experience Guide

Main Configuration Menu

Adding a New Model - Complete Wizard Flow

Step 1: Model Name

Step 2: Aliases Configuration

Step 3: Description

Step 4-5: Provider Configuration

Step 6-7: Configuration & Testing

Step 8-10: Optional Configuration

Model Management Interface

Listing Models

Generation Output Examples

Provider Configuration Examples

OpenAI Configuration

Claude (Anthropic) Configuration