godot
MCP server from tugcantopaloglu/godot-mcp
claude mcp add --transport stdio tugcantopaloglu-godot-mcp npx -y godot-mcp \ --env GMCP_DEBUG="false" \ --env GMCP_API_KEY="your-api-key-if-required"
How to use
Godot MCP - Full Control is a comprehensive MCP server that exposes a broad suite of Godot engine controls to an AI assistant. It provides 149 tools for runtime inspection, scene manipulation, project management, and runtime code execution within a running Godot project. Tools cover runtime node inspection, property reads/writes, dynamic scene modifications, signal management, animation and tween controls, file I/O, and even headless scene operations for parsing and editing .tscn scene files without starting a game. You can use these tools to script complex workflows, automate game testing, or build AI-assisted development workflows that interact with the Godot engine in real time. The server is built on TypeScript and extends the original godot-mcp, adding advanced capabilities like game_eval for executing GDScript and full node introspection via game_get_node_info.
How to install
Prerequisites:
- Node.js (v14+ recommended)
- npm or yarn
Step-by-step installation:
-
Create a workspace directory and navigate into it mkdir godot-mcp-workspace cd godot-mcp-workspace
-
Install the MCP server package from npm (assuming the package name is godot-mcp) npm install godot-mcp
or using yarn
yarn add godot-mcp
-
Run the MCP server (using npx integration as defined in mcp_config) npx -y godot-mcp
-
(Optional) Configure environment variables as needed GMCP_API_KEY=your-api-key GMCP_DEBUG=false npx -y godot-mcp
-
Verify the server starts and exposes the MCP endpoints
- Ensure your AI client can connect to the MCP server on the configured port.
If you prefer a local build or a different launcher, refer to the package.json scripts in the repository for alternative start commands.
Additional notes
Notes and tips:
- This MCP server exposes a large toolbox of Godot interactions; start with read-only checks (game_get_property, game_get_node_info) to understand the node trees before performing write operations.
- When using runtime code execution (game_eval), ensure you handle potential exceptions from GDScript execution and sandbox appropriately.
- Be mindful of performance: many tools (like listing all nodes or querying large scene trees) can be expensive on large projects.
- Use the headless scene operations (read_scene, modify_scene_node, attach_script) to inspect or modify scenes without launching the game.
- Environment variables: GMCP_API_KEY may be required for certain features; GMCP_DEBUG can help with verbose logs during setup.
- If you encounter connection issues, verify the MCP server is listening on the expected host/port and that the client uses compatible MCP protocol versions.
Related MCP Servers
mcp-for-beginners
This open-source curriculum introduces the fundamentals of Model Context Protocol (MCP) through real-world, cross-language examples in .NET, Java, TypeScript, JavaScript, Rust and Python. Designed for developers, it focuses on practical techniques for building modular, scalable, and secure AI workflows from session setup to service orchestration.
bitbucket
Bitbucket MCP - A Model Context Protocol (MCP) server for integrating with Bitbucket Cloud and Server APIs
mcp-arr
MCP server for *arr media management suite
crawlbase
Crawlbase MCP Server connects AI agents and LLMs with real-time web data. It powers Claude, Cursor, and Windsurf integrations with battle-tested web scraping, JavaScript rendering, and anti-bot protection enabling structured, live data inside your AI workflows.
unity-editor
An MCP server and client for LLMs to interact with Unity Projects
google-knowledge-graph
MCP server for Google's free public Knowledge Graph Search API - search for structured entity information about people, places, organizations, and concepts