This project provides an AI-powered code reader designed for learning and exchange, enabling users to interact with codebases using large language models. It supports both OpenAI's ChatGPT and self-hosted open-source LLMs, offering a novel way to understand and navigate code projects.

How It Works

The tool integrates with LLMs to process and analyze code repositories. Users can select between ChatGPT (GPT-3.5 or GPT-4) or local open-source models. For local models, users can extend functionality by creating custom LLM classes within the llms directory. The system leverages embeddings for local codebases, facilitating more context-aware interactions.

Quick Start & Requirements

• Install dependencies: pip install -r requirements.txt (for ChatGPT) or pip install -r llms/requirements_chatglm3.txt (for specific open-source models).
• Configure the .env file, a renamed copy of .env.tpl.
• Run the application: python main.py.
• Requires Python 3.10+.
• Recommended model context window: 16k+.

Highlighted Details

• Supports ChatGPT (gpt-3.5-turbo-1106 and gpt-4) and custom open-source LLMs.
• Allows integration of custom LLMs by inheriting from the LLM class.
• Emphasizes cleaning repositories by removing non-source files (e.g., archives, images) to prevent errors and manage API costs.

Maintenance & Community

The project is open for contributions and feedback, with the developer planning future updates.

Licensing & Compatibility

This project is strictly for learning and exchange and explicitly prohibits commercial use.

Limitations & Caveats

The project is noted as having many imperfections due to rapid development. Users are advised to monitor API usage and start with smaller projects to manage costs and potential errors.