godot-bridge -public
MCP server + Godot 4.x addon to control the Godot Editor via WebSocket/JSON-RPC
claude mcp add --transport stdio marcuzzifranco-godot-bridge-mcp-public python C:/RUTA/COMPLETA/mcp-server/src/main.py \ --env GODOT_WS_URL="ws://127.0.0.1:49631" \ --env GODOT_TOKEN_FILE="C:/Users/TU_USUARIO/AppData/Roaming/Godot/app_userdata/TU_PROYECTO/godotbridge_token.txt" \ --env GODOT_MCP_VERBOSE="1"
How to use
Godot MCP Bridge acts as a bridge between MCP-compatible AI agents and Godot 4.x, allowing you to manipulate the Godot editor, scenes, nodes, resources, and scripts directly from your AI workflow. The server exposes a set of MCP-enabled commands (organized by Project, Editor, Scenes, Nodes, Meshes/Materials, Sprites/Textures, Physics, Lighting/Cameras, Audio, Scripts, Execution, and Files) that let you query state, create or modify content, and trigger actions inside the editor. To use it, start the MCP server as described in the installation steps and ensure your MCP client (Cursor, Claude Desktop, Windsurf, or any compliant client) can reach the WebSocket endpoint exposed by the Godot editor plugin. Once connected, you can issue natural-language-like commands such as creating a new 3D scene, adding a node, configuring a camera, or writing a script, and the MCP bridge will translate them into Godot operations. The included toolset covers project information, editor state, scene management, node manipulation, mesh/material handling, lighting and audio, and script operations, enabling a wide range of editor automation from your preferred AI assistant.
How to install
Prerequisites:\n- Godot 4.x installed (tested with 4.4.x)\n- Python 3.9+ installed\n- Access to a compatible MCP client (Cursor, Claude Desktop, Windsurf, etc.)\n\nStep-by-step:\n1) Install the Godot Addon (per the Godot project instructions in the README) and verify the gateway port 49631 is active. The addon prints a token file for secure communication as described in the docs.\n2) Install the MCP server dependencies:\nbash\npip install websockets>=10.0\n\n3) Configure the MCP client to point at the server:\n- Create or edit your mcp.json (location varies by client, see the README) with a server entry like the example below.\njson\n{\n "mcpServers": {\n "godot": {\n "command": "python",\n "args": ["C:/RUTA/COMPLETA/mcp-server/src/main.py"],\n "env": {\n "GODOT_WS_URL": "ws://127.0.0.1:49631",\n "GODOT_TOKEN_FILE": "C:/Users/TU_USUARIO/AppData/Roaming/Godot/app_userdata/TU_PROYECTO/godotbridge_token.txt",\n "GODOT_MCP_VERBOSE": "1"\n }\n }\n }\n}\n\n4) Start Cursor or your MCP client and connect to ws://127.0.0.1:49631. Ensure the Godot plugin prints that the RPC server is listening on port 49631.\n5) Restart Cursor to load the new configuration and begin issuing commands to Godot.\n
Additional notes
Tips and caveats:\n- Ensure the token file path matches your system and that the Godot plugin has written the token. If you use the token file option, the token is refreshed automatically by the plugin.\n- If you choose the direct token option, you will need to update the token manually if it changes.\n- The MCP verbose flag is optional; set it to 1 to enable detailed logs for debugging.\n- The provided command assumes a local setup; adjust paths for Windows, macOS, or Linux as needed.\n- To use the built-in WebSocket server for custom clients, connect to ws://127.0.0.1:49631.\n- The MCP bridge supports a broad set of tools (scene management, node operations, materials, scripts, input, physics, audio, etc.). Explore the Tooling table in the README to discover exact commands and usage patterns.
Related MCP Servers
mcp-pinecone
Model Context Protocol server to allow for reading and writing from Pinecone. Rudimentary RAG
lc2mcp
Convert LangChain tools to FastMCP tools
Pare
Dev tools, optimized for agents. Structured, token-efficient MCP servers for git, test runners, npm, Docker, and more.
freecad
FreeCAD MCP - Open-source Model Context Protocol server for FreeCAD automation
skill-to
Convert AI Skills (Claude Skills format) to MCP server resources - Part of BioContextAI
cursor-feedback-extension
Save your Cursor monthly quota! Unlimited AI interactions in one conversation via MCP feedback loop.