Automata is an open-source project aiming to create a self-coding AI agent capable of autonomous software development. It targets AI researchers and developers interested in exploring emergent AI capabilities and AGI, offering a framework for AI to learn, document, and write code.

How It Works

Automata integrates Large Language Models (LLMs) with vector databases to enable code understanding and generation. It creates code embeddings and documentation to build a code graph, facilitating symbol dependency analysis. This approach allows the AI to iteratively improve its coding skills by learning from its own outputs and feedback, mimicking an autonomous craftsman.

Quick Start & Requirements

• Installation: Clone the repository, install dependencies via poetry install, and configure with poetry run automata configure. Docker installation is also available.
• Prerequisites: Requires Python 3.x, Poetry, and potentially C++ build tools and updated GCC/G++ on Windows. API keys for OpenAI and GitHub are necessary for full functionality.
• Resources: SCIP indices are required for search functionality, which may need manual generation.
• Docs: https://github.com/emrgnt-cmplxty/Automata

Highlighted Details

• Leverages LLMs (e.g., GPT-4) and vector databases for code comprehension and generation.
• Employs SCIP indexing for code graph creation and symbol dependency analysis.
• Supports building custom agents with specific tools and instructions.
• Includes examples for creating SymbolCodeEmbedding and SymbolDocEmbedding classes.

Maintenance & Community

• Uses GitHub issues for tracking and Automata Discussions for general questions.
• Adheres to a code of conduct.

Licensing & Compatibility

• Licensed under the Apache License 2.0.
• Permissive license suitable for commercial use and integration into closed-source projects.

Limitations & Caveats

The project is described as an evolving effort, with the ultimate goal of full autonomy being a long-term endeavor. Specific performance benchmarks or stability guarantees are not detailed.