Auto-Commit

AI-powered git commit message generator with intelligent diff processing and multi-provider support

According to git staged content, AI summarizes changes and generates commit messages following Conventional Commits format. Supports multiple LLM providers including OpenAI, Anthropic, Google, DeepSeek, Qwen, and Ollama.

✨ Features

• 🤖 Multi-LLM Support: OpenAI, Anthropic, Google, DeepSeek, Qwen, Ollama
• 📊 Intelligent Token Management: Precise token counting with @xenova/transformers
• 🗜️ Adaptive Diff Compression: 4-layer strategy to handle large diffs
• 🎯 Conventional Commits: Generates properly formatted commit messages
• ⚡ Dynamic Model Lists: Fetches latest available models from APIs
• 🔄 Smart Fallback: Graceful degradation when token limits are exceeded

🚀 Installation

npm install -g @whiteriverbay/auto-commit

Or use directly with npx:

npx @whiteriverbay/auto-commit

📖 Usage

Initial Setup

Configure your LLM provider and model:

auto-commit config

Interactive configuration process:

🔧 Auto-commit Configuration

? Select LLM Provider: › 
❯ OpenAI
  Anthropic (Claude)
  Google (Gemini)
  DeepSeek
  Qwen (通义千问)
  Ollama (Local)
  Exit

? Enter your OpenAI API key: ********
? Continue with configuration? Yes

Fetching available models...
? Select a model: ›
❯ gpt-4
  gpt-4-turbo  
  gpt-3.5-turbo
  Exit

✅ Configuration saved successfully!
Provider: OpenAI
Model: gpt-4

Generate Commit Messages

Stage your changes and let AI generate the commit message:

# Stage your changes
git add .

# Generate and commit
auto-commit

🎬 Demo

Basic Usage

$ auto-commit
🔍 Analyzing staged changes...
🤖 Generating commit message using AI...
Provider: google
Model: gemini-2.5-flash
Token limit: 50,000
Loading tokenizer: Xenova/gpt-4o
Compression level: full
Diff tokens: 1,245
✅ Generated commit message:
"feat(auth): add JWT token validation middleware"
? Do you want to commit with this message? Yes
📝 Committing changes...
✅ Successfully committed changes!
Commit message: "feat(auth): add JWT token validation middleware"

Large Diff with Compression

$ auto-commit
🔍 Analyzing staged changes...
🤖 Generating commit message using AI...
Provider: openai
Model: gpt-4
Token limit: 4,096
Loading tokenizer: Xenova/gpt-4o
Compression level: smart
Diff tokens: 3,891
✅ Generated commit message:
"feat(dashboard): implement user analytics with charts and filters

- Add ChartComponent with D3.js integration
- Implement user filtering by date range
- Add export functionality for analytics data"
? Do you want to commit with this message? Yes
📝 Committing changes...
✅ Successfully committed changes!

Extreme Compression for Very Large Changes

$ auto-commit
🔍 Analyzing staged changes...
🤖 Generating commit message using AI...
Provider: anthropic
Model: claude-3-5-sonnet-20241022
Token limit: 8,192
Loading tokenizer: Xenova/claude-tokenizer
Compression level: extreme
Diff tokens: 7,854
✅ Generated commit message:
"refactor: major codebase restructuring and dependency updates

Modified 47 files (+2,847 -1,923 lines)"
? Do you want to commit with this message? Yes
📝 Committing changes...
✅ Successfully committed changes!

Error Handling

$ auto-commit
⚠️  No staged changes found.
Use `git add <files>` to stage files before running auto-commit.

$ auto-commit
❌ Error: Not a git repository
Please run this command in a git repository directory.

🛠️ Supported Providers

| Provider | Status | Models | Token Limit | Dynamic List | |----------|--------|--------|-------------|--------------| | OpenAI | ✅ Active | GPT-4, GPT-3.5-turbo, etc. | 4,096 | ✅ | | Anthropic | ✅ Active | Claude 3/4 series | 8,192 | ✅ | | Google | ✅ Active | Gemini 1.5/2.0/2.5 series | 50,000 | ✅ | | DeepSeek | ✅ Active | DeepSeek Chat/Coder | 4,096 | ✅ | | Qwen | ✅ Active | Qwen Turbo/Plus/Max | 4,096 | 📋 Fixed | | Ollama | ✅ Active | Local models | 2,048 | ✅ |

✅ = Dynamic model fetching
📋 = Predefined model list

🧠 Intelligent Compression

Auto-commit uses a 4-layer compression strategy to handle diffs of any size:

1. Full Level (Optimal)

File: src/components/UserProfile.tsx
Changes:
+import { useState, useEffect } from 'react';
+import { User } from '../types/User';
+
+export const UserProfile = ({ userId }: { userId: string }) => {
+  const [user, setUser] = useState<User | null>(null);

2. Smart Level (Key Changes)

File: src/components/UserProfile.tsx [core] (+45 -12)
Key changes:
  + import { useState, useEffect } from 'react'
  + export const UserProfile = ({ userId }: { userId: string })
  + const [user, setUser] = useState<User | null>(null)

3. Minimal Level (Summary)

Modified: 3 core files, 2 config files (+127 -89 lines)

4. Extreme Level (Stats Only)

Modified 8 files (+234 -156 lines)

⚙️ Configuration

Configuration is stored in ~/.auto-commit/config.yml:

provider: google
apiKey: your-api-key-here
model: gemini-2.5-flash

Environment Variables

You can also set API keys via environment variables:

export OPENAI_API_KEY="your-key"
export ANTHROPIC_API_KEY="your-key" 
export GOOGLE_GENERATIVE_AI_API_KEY="your-key"

🔧 Development

Prerequisites

• Node.js 16+
• Git repository

Local Development

# Clone the repository
git clone https://github.com/ethanzhrepo/auto-commit.git
cd auto-commit

# Install dependencies
npm install

# Build the project
npm run build

# Run locally
npm run dev

# Or test built version
node dist/index.js

Testing

# Create test changes
echo "console.log('test');" > test.js
git add test.js

# Test the tool
npm run dev
# or
node dist/index.js

🤝 Contributing

1. Fork the repository
2. Create a feature branch: git checkout -b feature/amazing-feature
3. Commit your changes: git commit -m 'feat: add amazing feature'
4. Push to the branch: git push origin feature/amazing-feature
5. Open a Pull Request

📄 License

This project is licensed under the ISC License - see the LICENSE file for details.

🙏 Acknowledgments

• Vercel AI SDK - Unified AI provider interface
• Transformers.js - Client-side tokenization
• Conventional Commits - Commit message specification

🐛 Troubleshooting

Common Issues

"No configuration found"

auto-commit config

"Invalid JSON response"

• Usually resolved automatically with intelligent compression
• Try different model or provider if issues persist

"Token limit exceeded"

• The tool automatically compresses diffs to fit token limits
• No action needed - compression happens transparently

Model loading errors

• Check your API key is valid
• Verify internet connection for model downloads
• Some models may take time to download on first use

Getting Help

• 🐛 Report Issues
• 💬 Discussions
• 📧 Support: GitHub Issues