Wolverine provides "regenerative healing" for Python scripts by automatically fixing crashes using GPT-4. It's designed for developers who want to improve script resilience and reduce debugging time, especially when dealing with complex or numerous bugs.

How It Works

Wolverine executes a Python script, captures any exceptions and stack traces, and then sends this information to GPT-4. The model analyzes the error and generates code edits to fix the bug. Wolverine then applies these edits to the original script and reruns it, repeating the process until the script executes successfully. This iterative approach aims to resolve multiple bugs automatically.

Quick Start & Requirements

• Install: pip install -r requirements.txt
• Prerequisites: Python 3.x, OpenAI API key (add to .env file).
• Usage: python -m wolverine examples/buggy_script.py "subtract" 20 3
• More info: Twitter demo video

Highlighted Details

• Automatically reruns scripts until fixed.
• Supports GPT-4 (default) and GPT-3.5-turbo.
• Optional --confirm=True flag for manual approval of edits.

Maintenance & Community

This project is marked as DEPRECATED and suggests trying Mentat instead. Future plans mentioned in the README are unlikely to be implemented.

Licensing & Compatibility

The README does not specify a license.

Limitations & Caveats

The project is deprecated. It may struggle with code indentation issues and is limited to Python scripts. Handling large files or multiple codebases is not explicitly addressed.