Skip to content

Godot MCP

An MCP for Godot that lets you create and edit games in the Godot game engine with tools like Claude

Install

$git clone https://github.com/ee0pdt/Godot-MCP

Install command taken from the project's README or registry entry. Always review code before running it.

Godot MCP (Model Context Protocol)

A comprehensive integration between Godot Engine and AI assistants using the Model Context Protocol (MCP). This plugin allows AI assistants to interact with your Godot projects, providing powerful capabilities for code assistance, scene manipulation, and project management.

Features

  • Full Godot Project Access: AI assistants can access and modify scripts, scenes, nodes, and project resources
  • Two-way Communication: Send project data to AI and apply suggested changes directly in the editor
  • Command Categories:
    • Node Commands: Create, modify, and manage nodes in your scenes
    • Script Commands: Edit, analyze, and create GDScript files
    • Scene Commands: Manipulate scenes and their structure
    • Project Commands: Access project settings and resources
    • Editor Commands: Control various editor functionality

Quick Setup

1. Clone the Repository

git clone https://github.com/ee0pdt/godot-mcp.git
cd godot-mcp

2. Set Up the MCP Server

cd server
npm install
npm run build
# Return to project root
cd ..

3. Set Up Claude Desktop

  1. Edit or create the Claude Desktop config file:

    # For macOS
    nano ~/Library/Application\ Support/Claude/claude_desktop_config.json
    
  2. Add the following configuration (or use the included claude_desktop_config.json as a reference):

    {
      "mcpServers": {
        "godot-mcp": {
          "command": "node",
          "args": [
            "PATH_TO_YOUR_PROJECT/server/dist/index.js"
          ],
          "env": {
            "MCP_TRANSPORT": "stdio"
          }
        }
      }
    }
    

    Note: Replace PATH_TO_YOUR_PROJECT with the absolute path to where you have this repository stored.

  3. Restart Claude Desktop

4. Open the Example Project in Godot

  1. Open Godot Engine
  2. Select "Import" and navigate to the cloned repository
  3. Open the project.godot file
  4. The MCP plugin is already enabled in this example project

Using MCP with Claude

After setup, you can work with your Godot project directly from Claude using natural language. Here are some examples:

Example Prompts

@mcp godot-mcp read godot://script/current

I need help optimizing my player movement code. Can you suggest improvements?
@mcp godot-mcp run get-scene-tree

Add a cube in the middle of the scene and then make a camera that is looking at the cube.
@mcp godot-mcp read godot://scene/current

Create an enemy AI that patrols between waypoints and attacks the player when in range.

Natural Language Tasks Claude Can Perform

  • "Create a main menu with play, options, and quit buttons"
  • "Add collision detection to the player character"
  • "Implement a day/night cycle system"
  • "Refactor this code to use signals instead of direct references"
  • "Debug why my player character falls through the floor sometimes"

Available Resources and Commands

Resource Endpoints:

  • godot://script/current - The currently open script
  • godot://scene/current - The currently open scene
  • godot://project/info - Project metadata and settings

Command Categories:

Node Commands

  • get-scene-tree - Returns the scene tree structure
  • get-node-properties - Gets properties of a specific node
  • create-node - Creates a new node
  • delete-node - Deletes a node
  • modify-node - Updates node properties

Similar plugs

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.

Java 面试 & 后端通用面试指南,覆盖计算机基础、数据库、分布式、高并发、系统设计与 AI 应用开发

MCP serverDatabases+24 this week

Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype…

Explore related