This repository provides a free, open-source implementation of autonomous AI agents like Auto-GPT and BabyAGI, targeting users who want to experiment with AI without incurring costs from paid APIs. It offers a simplified approach to autonomous task execution, democratizing access to advanced AI capabilities for individuals and small businesses.

How It Works

The project achieves its goal by reverse-engineering ChatGPT and other popular LLM services (HuggingChat, Bing Chat, Google Bard) to create custom Langchain wrappers. This allows users to leverage these LLMs without API keys, relying instead on browser cookies and Hugging Face tokens. The core innovation lies in its "plug-and-play" LLM wrapper, enabling integration with existing Langchain functionalities and tools.

Quick Start & Requirements

• Installation: Via Dev Container in VSCode or python3 -m pip install -r requirements.txt.
• Prerequisites: Docker Desktop (for Dev Container), VSCode with Dev Container extension. Requires obtaining browser cookies/tokens for ChatGPT, HuggingChat, Bing Chat, or Google Bard.
• Usage: Run python BABYAGI.py or python AUTOGPT.py. A Colab notebook is also available for immediate use.
• Links: Colab Notebook, Video Demo

Highlighted Details

• Supports multiple free LLM backends: ChatGPT (via reverse-engineered API), HuggingChat, Bing Chat, and Google Bard.
• Includes pre-built agents for BabyAGI, Auto-GPT, and custom agents with internet search, Python REPL, and Wikipedia access.
• Offers a custom LLM wrapper for Langchain, facilitating easy integration.
• Provides a VSCode Dev Container setup for local development and execution.

Maintenance & Community

The project actively seeks contributions and pull requests. Links to community resources like Discord/Slack are not explicitly provided in the README.

Licensing & Compatibility

The repository is licensed under the MIT License. The project explicitly states it is for "educational purposes only" and is not affiliated with or endorsed by API providers. Users are advised to follow the Terms of Service of each website they use cookies from.

Limitations & Caveats

The project relies on reverse-engineered APIs, which are subject to change or discontinuation by the respective service providers. Usage of browser cookies may violate the Terms of Service of these providers. The README notes a 50 requests/hour limit for the reverse-engineered ChatGPT API.