Git Diff AI

Git Diff AI is a powerful tool that generates commit messages based on your staged changes using AI. It analyzes your git diff and produces both individual file commit messages and a summary commit message, making your git workflow more efficient and descriptive.

Description

This tool uses Ollama's AI models to analyze git diffs and generate meaningful commit messages. It processes each changed file individually and then creates an overall summary, helping developers create more descriptive and consistent commit messages with minimal effort.

Dependencies

• Node.js (version 18 or higher)
• Ollama CLI (installed and configured on your system)

Checking Dependencies

1. Check your Node.js version:

node -v

2. Check if Ollama is installed:

ollama --help

If you see the help output, Ollama is installed correctly.

Installing Ollama

If Ollama is not installed, follow these steps:

1. Visit the Ollama installation page.
2. Download and install the appropriate version for your operating system.
3. After installation, run ollama --help to verify the installation.

Installation

1. Clone this repository:

    git clone https://github.com/yourusername/git-diff-ai.git
    cd git-diff-ai

2. Install the dependencies:

npm install

Usage

Run the script using npm:

npm run git-diff-ai -- [options]

Options

• --model <model-name>: Specify the Ollama model to use (default: qwen2.5:3b)
• --max-diff-size <size>: Maximum size of diff to process in characters (default: 10000)
• --max-files <number>: Maximum number of files to process (default: 50)

Examples

1. Using default settings:

npm run git-diff-ai

2. Using a specific model and increased diff size(model has to be pulled with ollama before):

npm run git-diff-ai -- --model gpt-4 --max-diff-size 50000

3. Processing more files:

npm run git-diff-ai -- --max-files 100

How It Works

1. The tool retrieves the git diff for staged changes.
2. It processes each file diff individually, skipping excluded file types and large diffs.
3. For each valid diff, it generates a commit message using the specified AI model.
4. After processing all files, it generates a summary commit message.
5. The individual file commit messages and the summary message are displayed in the console.

General Recommendations and Explanations

1. Hardware Considerations: This tool uses your local hardware for computations. Choose an appropriate model based on your machine's capabilities. Larger models may provide better results but require more computational power and time.

2. Data Privacy: All processing is done locally on your machine. No data is stored or sent to external servers, ensuring the privacy and security of your code.

3. Git Best Practices:

• Avoid staging too many files for a single commit. This not only increases computation time but also goes against the recommended git workflow.
• Aim for atomic commits: each commit should represent a single logical change. This makes your git history more meaningful and easier to understand. Learn more about atomic commits.

4. Model Selection: Start with smaller models and gradually move to larger ones if needed. The default model (qwen2.5:3b) offers a good balance between performance and quality for most use cases.

5. Performance: If you're experiencing slow performance, consider:

• Using a smaller model
• Reducing the --max-diff-size or --max-files parameters
• Staging fewer files for each commit

6. Review Generated Messages: While this tool aims to generate meaningful commit messages, always review and adjust them as necessary to ensure they accurately describe your changes.

Troubleshooting

1. Error: Ollama is not installed or configured

• Make sure Ollama is installed on your system and properly configured.
• Check if you can run Ollama commands directly in your terminal.

2. Error: No changes detected

• Ensure you have staged changes in your git repository.
• Run git status to verify that there are staged changes.

3. Error: Model not found

• Verify that the specified model is available in your Ollama installation.
• Try using a different model or the default model.

4. Error: Maximum diff size exceeded

• If you're working with large files, try increasing the --max-diff-size parameter.

5. Error: Too many files

• If you're committing changes to many files, try increasing the --max-files parameter.

6. Performance issues or slow responses

• Consider using a smaller or faster model.
• Reduce the --max-diff-size or --max-files parameters.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

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