Godcoder is a local-first, open-source AI coding agent designed for desktop use, prioritizing user privacy by keeping code on the machine and only sending requests to user-provided LLM keys. It offers advanced capabilities beyond simple code editing, including autonomous agent harness development and GUI automation, making it suitable for developers and power users seeking a private, powerful AI coding assistant.

How It Works

Godcoder runs as a native desktop application, connecting directly to user-configured LLM providers (OpenAI, Anthropic, or compatible APIs) without intermediate cloud backends. Its defining feature is "Harness mode," where the AI autonomously scaffolds, engineers, and optimizes its own agent harness and toolset in real-time, creating a self-improving loop. Additionally, "CoWork mode" allows the agent to learn and execute human-action tasks, such as GUI interactions and OS scripting, by driving the Open Cowork desktop app. Both modes employ a self-optimizing cycle that compounds learning and improves performance over time.

Quick Start & Requirements

• Install: Build from source. Navigate to apps/desktop and run npm install followed by npm run tauri:build for production, or npm run tauri:dev for development. A Windows shortcut (launch-godcoder.bat) is provided for easier setup.
• Prerequisites: Rust (stable), Tauri 2 system prerequisites for your OS, Node.js 20+ and npm (optional, for Context Engine), Docker with Compose (for Context Engine).
• Setup: On first launch, configure LLM provider details (base_url, api_key, model) in Settings. To enable the optional Context Engine, set up the Go indexing service via Docker (services/context-engine/README.md) and enable it in the app settings.
• Links: ARCHITECTURE.md, CONTRIBUTING.md, Context Engine Setup.

Highlighted Details

• Real-Time Self-Built Harness: The agent autonomously creates and refines its own operational harness without human prompting.
• Multi-Mode Operation: Supports Ask, Plan, Coding, Freestyle, Harness, and CoWork modes for diverse AI-driven tasks.
• LLM Provider Agnostic: Integrates with OpenAI, Anthropic, and any OpenAI-compatible API endpoints directly.
• GUI/OS Automation: CoWork mode enables the agent to learn and execute human-action tasks via graphical user interface automation.
• Local-First Privacy: All user code and data remain on the local machine, with no vendor backend transit.

Maintenance & Community

Contributions are welcomed via CONTRIBUTING.md. Bug reports should be filed as GitHub issues, and ideas can be discussed in the project's discussions section. No specific community channels (e.g., Discord, Slack) or notable contributors/sponsors are detailed in the provided README.

Licensing & Compatibility

The project is licensed under the MIT License, permitting commercial use and integration with closed-source projects.

Limitations & Caveats

Prebuilt release binaries and installers are not yet available, requiring users to build from source. The advanced Context Engine feature necessitates a separate setup process using Docker. Some vendored source builds may emit platform-specific warnings during the bootstrap process, though these are treated as optional to ensure completion.