This project addresses the challenge of automating complex software engineering tasks by deploying a swarm of AI agents to analyze and modify codebases without explicit human prompts. It targets engineers, researchers, and power users seeking to accelerate development cycles and resolve codebase issues autonomously. The primary benefit is the potential for significant time savings and improved code quality through parallel, AI-driven problem-solving.

How It Works

Full Self Coding (FSC) employs a multi-agent architecture where numerous AI agents (initially Claude Code and Gemini CLI) operate concurrently within isolated Docker containers. The system analyzes a given codebase, decomposes identified problems into discrete tasks, and assigns these tasks to available agents. Core components include a ConfigReader for managing settings, DockerInstance for container lifecycle management, TaskSolver for executing agent tasks, and an Analyzer for initial codebase assessment and task generation. This approach leverages containerization for consistent, secure execution and parallel processing for rapid iteration and problem resolution.

Quick Start & Requirements

• Prerequisites: Node.js (v18.0.0+), Docker (latest), Git, Bun (recommended).
• Installation: Install the package globally using Bun: bun install -g full-self-coding.
• Execution: Clone a target repository (git clone <repo_url>), navigate into it (cd repo), and run the tool with the repository URL: full-self-coding <git-repository-url> or node dist/main.js <git-repository-url>.
• Documentation: Usage examples and configuration details are provided within the README.

Highlighted Details

• Multi-Agent System: Integrates multiple AI models (Claude Code, Gemini CLI, planned Codex) for diverse problem-solving capabilities.
• Containerized Execution: Utilizes Docker for secure, isolated, and reproducible task environments.
• Intelligent Codebase Analysis: Automatically analyzes code to identify and decompose tasks for AI agents.
• Parallel Processing: Executes multiple AI agents concurrently across numerous Docker containers, managed for resource efficiency.
• Automated Code Modification: Generates and applies code changes directly to the codebase, with git diff tracking.

Maintenance & Community

The project is developed by the "Full Self Coding team." Support and feature requests are managed via GitHub Issues, and community discussions can occur on GitHub Discussions. Specific contributor details, sponsorships, or active community channels (like Discord/Slack) are not detailed in the provided README.

Licensing & Compatibility

The project is licensed under the MIT License. This license is permissive and generally allows for commercial use, modification, and distribution, including within closed-source applications.

Limitations & Caveats

The system's effectiveness is dependent on the capabilities and availability of external AI APIs (Anthropic, Google), potentially incurring costs and rate limits. The "node:latest" Docker image may not be suitable for all project types. While designed for parallel execution, managing a large number of agents (100-1000) could present significant operational complexity and resource demands. The integration of certain agents, like OpenAI Codex, is noted as planned, indicating it may not be fully implemented.