tugcantopaloglu/godot-mcp

What it solves

Godot MCP allows AI assistants to programmatically control the Godot game engine. It bridges the gap between an LLM's ability to write code and the actual execution, inspection, and modification of a game project, enabling AI to build, debug, and test games in real-time.

How it works

The project implements a Model Context Protocol (MCP) server that exposes 157 specialized tools to an AI assistant. These tools allow the AI to interact with Godot through two primary modes:

1. Headless Scene Operations: Modifying .tscn files, creating resources, and managing project settings without needing the game to be running.
2. Runtime Interaction: Using a TCP-based server to execute GDScript code (game_eval), inspect node properties, manipulate the scene tree, and simulate user input (keyboard, mouse, gamepad) while the game is active.

It supports both GDScript and .NET/C# projects, providing tools for script validation and project scaffolding.

Who it’s for

Game developers using Godot 4.4+ who want to leverage AI agents to automate game creation, perform runtime debugging, or generate complex scene structures and logic.

Highlights

• Extensive Toolset: 157 tools covering everything from 3D/2D rendering, physics, and audio to networking and UI controls.
• Runtime Execution: Ability to execute arbitrary GDScript code and await async operations while the game is running.
• Deep Introspection: Full node inspection, property reading/writing, and signal management at runtime.
• C# Support: First-class support for scaffolding .NET projects and generating C# scripts.
• Headless Editing: Ability to parse and modify scene files and project settings without launching the editor.
• Integrated Diagnostics: Tools to validate GDScript syntax and type errors headlessly.