cmai 0.1.2

AI Powered Commit Message Normalization Tool

CMAI - AI-Powered Commit Message Normalizer

CMAI is an intelligent command-line tool that leverages AI to transform informal or colloquial commit messages into standardized, professional Git commit messages. It analyzes your staged changes and uses advanced language models to generate clear, concise, and conventional commit messages.

🌟 Features

• AI-Powered Normalization: Converts informal commit descriptions into professional, standardized commit messages
• Git Integration: Automatically analyzes staged changes to provide context-aware suggestions
• Multiple AI Provider Support: Supports various AI providers including OpenAI-compatible APIs, Bailian (Qwen), DeepSeek, SiliconFlow, and local Ollama models
• Extensible Architecture: Easy to add new AI providers through the provider factory system
• Configurable: Customizable prompt templates and model settings
• Token Usage Tracking: Monitor AI token consumption for cost management
• Comprehensive Logging: Detailed logging for debugging and monitoring with stream output support

🚀 Quick Start

Installation

Install CMAI using pip:

pip install cmai

Or install from source:

git clone https://github.com/yumuzhihan/cmai.git
cd cmai
pip install -e .

Quick Setup

1. Configure your AI provider (this step is required):

Create ~/.config/cmai/settings.env:

# Example: Using OpenAI
PROVIDER=openai
API_KEY=your_openai_api_key
MODEL=gpt-4o-mini

# Or using Ollama locally
# PROVIDER=ollama
# OLLAMA_HOST=http://localhost:11434
# MODEL=qwen2.5:7b

2. Set your API key (if using remote providers):

# For OpenAI
export OPENAI_API_KEY=your_api_key

# For Bailian
export DASHSCOPE_API_KEY=your_api_key

# For DeepSeek
export DEEPSEEK_API_KEY=your_api_key

3. Test the installation:

# Stage some changes
git add .

# Generate a commit message
cmai "fixed a bug"

Basic Usage

1. Stage your changes in Git:

git add .

2. Use CMAI to generate a normalized commit message:

cmai "fix some bugs in user authentication"

3. The tool will output a standardized commit message like:

Commit message: Fix authentication bugs in user login module
Tokens used: 45

⚠️ Important Setup Reminder: Before using CMAI, you must configure at least two settings:

• PROVIDER: The AI provider you want to use (e.g., openai, bailian, ollama)
• MODEL: The specific model name (this is required for all providers)

Without these configurations, CMAI will fail to run. See the Configuration section for details.

🔧 Configuration

CMAI uses a configuration file located at ~/.config/cmai/settings.env. The configuration file will be automatically created on first run.

Environment Variables

Create or edit ~/.config/cmai/settings.env:

# AI Provider Configuration
PROVIDER=openai
API_BASE=https://api.openai.com/v1
API_KEY=your_api_key_here
MODEL=gpt-4o-mini

# For Bailian (Qwen) API
# PROVIDER=bailian
# API_BASE=https://dashscope.aliyuncs.com/compatible-mode/v1
# API_KEY=your_dashscope_api_key
# MODEL=qwen-turbo-latest

# For DeepSeek API
# PROVIDER=deepseek
# API_BASE=https://api.deepseek.com/v1
# API_KEY=your_deepseek_api_key
# MODEL=deepseek-chat

# For SiliconFlow API
# PROVIDER=siliconflow
# API_BASE=https://api.siliconflow.cn/v1
# API_KEY=your_siliconflow_api_key
# MODEL=Qwen/Qwen2.5-7B-Instruct

# For Ollama (local)
# PROVIDER=ollama
# OLLAMA_HOST=http://localhost:11434
# MODEL=qwen2.5:7b

# Logging Configuration
LOG_LEVEL=INFO
LOG_FILE_PATH=/path/to/logfile.log

# Prompt Template (optional customization)
PROMPT_TEMPLATE=Please generate a standardized commit message based on the user description: {user_input}. The changes include: {diff_content}. Respond only with the normalized commit message in English.

Supported AI Providers

CMAI supports multiple AI providers through its extensible architecture:

1. OpenAI-Compatible APIs

• OpenAI: Official OpenAI API
• Bailian (Qwen): Alibaba Cloud's Qwen models
• DeepSeek: DeepSeek's AI models
• SiliconFlow: SiliconFlow's AI services
• ChatGPT: OpenAI ChatGPT models

2. Local Models

• Ollama: Run models locally using Ollama

Important: Model Configuration

⚠️ Important: You must specify a MODEL in your configuration. CMAI requires an explicit model name to function properly. Examples:

• OpenAI: gpt-4o-mini, gpt-4o, gpt-3.5-turbo
• Bailian: qwen-turbo-latest, qwen-plus-latest, qwen-max-latest
• DeepSeek: deepseek-chat, deepseek-coder
• SiliconFlow: Qwen/Qwen2.5-7B-Instruct, deepseek-ai/DeepSeek-V2.5
• Ollama: qwen2.5:7b, llama3.1:8b, codellama:7b

API Key Setup

Different providers require different API keys:

1. OpenAI: Set OPENAI_API_KEY or CMAI_API_KEY environment variable

export OPENAI_API_KEY=your_openai_api_key

2. Bailian (Qwen): Set DASHSCOPE_API_KEY environment variable

export DASHSCOPE_API_KEY=your_dashscope_api_key

3. DeepSeek: Set DEEPSEEK_API_KEY or CMAI_API_KEY environment variable

export DEEPSEEK_API_KEY=your_deepseek_api_key

4. SiliconFlow: Set SILICONFLOW_API_KEY or CMAI_API_KEY environment variable

export SILICONFLOW_API_KEY=your_siliconflow_api_key

5. Ollama: No API key required, but ensure Ollama is running locally

6. Configuration file: Add API_KEY=your_api_key_here to ~/.config/cmai/settings.env

7. Custom config file: Use the --config option to specify a different configuration file

📖 Usage Examples

Usage Examples

# Simple commit message normalization (uses configuration file settings)
cmai "updated readme file"

# Using a custom configuration file
cmai "fixed authentication bug" --config /path/to/custom/config.env

# Specifying a different repository
cmai "refactored user service" --repo /path/to/repo

Using Custom Configuration

# Use a specific configuration file
cmai "refactored auth system" --config /path/to/custom/config.env

Specifying Repository Path

# Analyze changes in a specific repository
cmai "fixed login bug" --repo /path/to/your/repo

Full Command Options

cmai [OPTIONS] MESSAGE

Arguments:
  MESSAGE  The informal commit message to be normalized [required]

Options:
  -c, --config TEXT  Path to configuration file
  -r, --repo TEXT    Git repository path
  --help            Show this message and exit

Note: Provider and model selection is configured through the configuration file or environment variables, not command-line arguments.

Configuration Examples for Different Providers

To use different AI providers, configure your ~/.config/cmai/settings.env file:

OpenAI

PROVIDER=openai
API_BASE=https://api.openai.com/v1
API_KEY=your_openai_api_key
MODEL=gpt-4o-mini

Bailian (Qwen)

PROVIDER=bailian
API_BASE=https://dashscope.aliyuncs.com/compatible-mode/v1
API_KEY=your_dashscope_api_key
MODEL=qwen-turbo-latest

DeepSeek

PROVIDER=deepseek
API_BASE=https://api.deepseek.com/v1
API_KEY=your_deepseek_api_key
MODEL=deepseek-chat

SiliconFlow

PROVIDER=siliconflow
API_BASE=https://api.siliconflow.cn/v1
API_KEY=your_siliconflow_api_key
MODEL=Qwen/Qwen2.5-7B-Instruct

Ollama (Local)

PROVIDER=ollama
OLLAMA_HOST=http://localhost:11434
MODEL=qwen2.5:7b

After configuring your preferred provider, simply use:

cmai "your commit message"

🏗️ Architecture

CMAI follows a modular architecture with the following components:

Core Components

• cmai.main: Entry point and CLI interface using Click
• cmai.core.normalizer: Core logic for commit message normalization
• cmai.core.get_logger: Logging factory and configuration with stream support
• cmai.config.settings: Configuration management using Pydantic

Provider System

• cmai.providers.base: Abstract base class for AI providers
• cmai.providers.openai_provider: OpenAI-compatible API implementation
• cmai.providers.ollama_provider: Ollama local model implementation
• cmai.providers.provider_factory: Factory for creating and managing providers
• cmai.providers.bailian_provider: Legacy Bailian provider (deprecated)

Utilities

• cmai.utils.git_staged_analyzer: Git repository analysis and diff extraction

Data Models

class AIResponse(BaseModel):
    content: str           # The normalized commit message
    model: str            # AI model used
    provider: str         # AI provider name
    tokens_used: Optional[int]  # Token consumption

Provider Factory System

CMAI uses a factory pattern for managing AI providers:

from cmai.providers.provider_factory import create_provider

# Create provider with default settings
provider = create_provider()

# Create specific provider with model
provider = create_provider("openai", model="gpt-4o-mini")

# Create Ollama provider
provider = create_provider("ollama", model="qwen2.5:7b")

🔌 Extending CMAI

Adding New AI Providers

To add support for a new AI provider, create a class that inherits from BaseAIClient and register it with the provider factory:

from cmai.providers.base import BaseAIClient, AIResponse
from cmai.providers.provider_factory import register_custom_provider

class CustomProvider(BaseAIClient):
    async def normalize_commit(self, prompt: str, **kwargs) -> AIResponse:
        # Implement your provider logic here
        # Must return AIResponse with content, model, provider, and tokens_used
        pass
    
    def validate_config(self) -> bool:
        # Implement configuration validation
        return True

# Register the provider
register_custom_provider("custom", CustomProvider)

Using the Provider Factory

The provider factory automatically manages different AI providers:

from cmai.providers.provider_factory import (
    create_provider,
    list_available_providers,
    register_custom_provider
)

# List all available providers
providers = list_available_providers()
print(providers)
# Output: {'openai': 'OpenaiProvider', 'ollama': 'OllamaProvider', ...}

# Create provider instances
openai_provider = create_provider("openai", model="gpt-4o-mini")
ollama_provider = create_provider("ollama", model="qwen2.5:7b")

Custom Prompt Templates

You can customize the prompt template by modifying the PROMPT_TEMPLATE setting:

PROMPT_TEMPLATE=Generate a standardized commit message based on: {user_input}. Changes: {diff_content}. Use conventional commit format.

Available placeholders:

• {user_input}: The user's informal commit message
• {diff_content}: Git diff information from staged changes

🛠️ Development

Setting Up Development Environment

1. Clone the repository:

git clone https://github.com/yumuzhihan/cmai.git
cd cmai

2. Create a virtual environment:

uv venv

# Activate virtual environment
# On Unix/macOS:
source .venv/bin/activate
# On Windows:
# .venv\Scripts\activate

3. Install development dependencies:

# Install all dependencies including dev dependencies
uv sync --dev

# Or if you prefer to install only production dependencies:
uv sync

Project Structure

cmai/
├── cmai/
│   ├── __init__.py
│   ├── main.py              # CLI entry point
│   ├── config/
│   │   ├── __init__.py
│   │   └── settings.py      # Configuration management
│   ├── core/
│   │   ├── __init__.py
│   │   ├── get_logger.py    # Logging utilities
│   │   └── normalizer.py    # Core normalization logic
│   ├── providers/
│   │   ├── __init__.py
│   │   ├── base.py          # Abstract provider interface
│   │   ├── openai_provider.py    # OpenAI-compatible implementation
│   │   ├── ollama_provider.py    # Ollama local model implementation
│   │   ├── provider_factory.py   # Provider factory system
│   │   └── bailian_provider.py   # Legacy Bailian provider
│   └── utils/
│       ├── __init__.py
│       └── git_staged_analyzer.py  # Git utilities
├── tests/                   # Test suite
├── docs/                    # Documentation
├── scripts/                 # Utility scripts
├── pyproject.toml          # Project configuration
├── uv.lock                 # UV lock file
├── LICENSE                 # AGPL-3.0 License
└── README.md              # This file

Running Tests

python -m pytest tests/

📋 Requirements

• Python 3.10 or higher
• Git (for repository analysis)
• Internet connection (for remote AI provider APIs)
• Ollama installation (for local AI models)

Dependencies

• click>=8.2.1 - Command-line interface
• openai>=1.91.0 - OpenAI-compatible API client
• ollama>=0.5.1 - Ollama Python client for local models
• pydantic>=2.11.7 - Data validation and settings
• pydantic-settings>=2.10.0 - Settings management

🤝 Contributing

We welcome contributions! Please follow these steps:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Make your changes
4. Add tests for new functionality
5. Commit your changes (git commit -m 'Add amazing feature')
6. Push to the branch (git push origin feature/amazing-feature)
7. Open a Pull Request

Contribution Guidelines

• Follow PEP 8 coding standards
• Add type hints to all functions
• Write comprehensive tests
• Update documentation for new features
• Ensure backward compatibility
• When adding new providers, register them in the provider factory

Adding New Providers

When contributing new AI providers:

1. Create a new file in cmai/providers/ (e.g., custom_provider.py)
2. Implement the BaseAIClient interface
3. Register the provider in provider_factory.py
4. Add configuration examples to the README
5. Include tests for the new provider

📄 License

This project is licensed under the GNU Affero General Public License v3.0 (AGPL-3.0). See the LICENSE file for details.

🙏 Acknowledgments

• Built with Click for the CLI interface
• Uses Pydantic for configuration and data validation
• Supports multiple AI providers including OpenAI, Bailian, DeepSeek, SiliconFlow, and Ollama
• Inspired by conventional commit standards
• Uses UV for fast Python package management

📞 Support

If you encounter any issues or have questions:

1. Check the Issues page
2. Create a new issue with detailed information about your setup (provider, model, configuration)
3. Review the documentation and configuration guide
4. For provider-specific issues, include your provider and model information

Common Issues

• "Model must be specified": Ensure you have set the MODEL in your configuration file or passed it as a command-line argument
• API key errors: Verify that your API key is correctly set for your chosen provider
• Connection errors: Check your internet connection and API endpoint URLs
• Ollama connection issues: Ensure Ollama is running locally and accessible at the configured host

---

Note: This tool requires access to AI language models. Please ensure you have appropriate API keys and understand the associated costs before using CMAI in production environments. For local usage, consider using Ollama with open-source models.