AI review pre commit

This project implements a custom pre-commit hook for pre-commit framework that checks the content of files staged for commit. The hook uses OpenAI's API to review the code and provide feedback before the commit is finalized. By default, it will fail if the AI returns feedback.

Project Structure

ai-review-pre-commit
├── .env
├── .gitignore
├── .pre-commit-hooks.yaml
├── main.py
├── pyproject.toml
├── README.md
├── requirements.txt
├── tests
│   └── test_utils.py
└── utils
    ├── __init__.py
    └── openai_consumer.py

Setup Instructions

1. Clone the repository:

   git clone [email protected]:jlariza/ai-review-pre-commit.git
   cd ai-review-pre-commit
   

2. Install the required dependencies:

   pip install -r requirements.txt
   

   Note: It is recommended to use a virtual environment to install the dependencies. You can create and activate a virtual environment with the following commands:

   python -m venv venv
   source venv/bin/activate  # On Windows, use `venv\Scripts\activate`
   

3. Set up your OpenAI API key:

   • Create OPENAI_API_KEY environment variable in the following format:

     OPENAI_API_KEY=your-api-key-here
     

Usage

CAUTION: This hook will send a request to the AI API per file changed per commit created. For most API's, like OpenAI, each call has an associated pricing. Make sure to review your project's budget before integrating this project in your workflow.

By default, the hook will only ask for a general review to the selected AI Model. However, this may be extended. Please, check configuration for more.

Local usage

When you attempt to make a commit, the pre-commit hook will automatically run. It will:

1. Retrieve the staged changes using git diff --staged.
2. Send the changes to OpenAI's API for review.
3. Provide feedback on the code. If no issues are found, the commit will proceed. Otherwise, the commit will be aborted, and feedback will be displayed.

To manually test the functionality, you can run the try-repo script:

pre-commit try-repo .

Extenal usage

If you want to use this functionality on an external repo:

1. Ensure pre-commit is installed
2. Ensure the OPENAI_API_KEY exists on your environment
3. Add the file .pre-commit-config.yaml on your repository's root directory, with the following configuration:

repos:
-   repo: https://github.com/jlariza/ai-review-pre-commit
    rev: 0.0.1
    hooks:
    -   id: ai-review
        args: [...]

4. run pre-commit install to set up the git hook scripts
5. Commit away!

configuration

ai-review hooks allows three different arguments:

• --format: If this arg is added, an extra format review will be required to the API. It will add one call per file per commit created to the flow.
• --security: If this arg is added, an extra security review will be required to the API, based in OWASP. It will add one call per file per commit created to the flow.
• --no-fail: If this arg is added, the hook will never fail; even if the AI returned feedback.

Testing

To run the tests for the utility functions, navigate to the tests directory and execute:

pytest tests/test_utils.py

This will ensure that the utility functions behave as expected and meet the defined criteria.

Disclaimer

This project is a work in progress and should not be used in production. The functionality and reliability of the code are still under development, and there may be bugs or incomplete features.

Contributing

Contributions are welcome! Please open an issue or submit a pull request for any enhancements or bug fixes.

TODO list

• Add support for additional AI models extending AIConsumerProtocol protocol.
• Improve feedback filterting to avoid non relevant blocking feedback.
• Optimize API call efficiency to reduce costs.
• Improve error handling and logging.
• Create detailed documentation for configuration options.