Skip to content

Godot MCP Pro

162 MCP tools for AI-powered Godot 4 development. Scene, animation, 3D, physics, particles, audio, shader, input simulation, runtime analysis, navigation, testing & more. $15…

Install

$git clone https://github.com/youichi-uda/godot-mcp-pro

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

Godot MCP Pro

Premium MCP (Model Context Protocol) server for AI-powered Godot game development. Connects AI assistants like Claude directly to your Godot editor with 187 powerful tools.

Architecture

AI Assistant ←—stdio/MCP—→ Node.js Server ←—WebSocket:6505—→ Godot Editor Plugin
  • Real-time: WebSocket connection means instant feedback, no file polling
  • Editor Integration: Full access to Godot's editor API, UndoRedo system, and scene tree
  • JSON-RPC 2.0: Standard protocol with proper error codes and suggestions

What's in this repo

⚠️ This public repo only contains the free Godot addon/plugin. The MCP server (Node.js, required to connect AI assistants) is distributed as part of the paid package — one-time purchase, lifetime updates:

The paid zip includes the addon, the server/ directory with pre-built JavaScript, INSTALL.md, and AI-client instructions. If you cloned this repo and don't see a server/ folder, that's expected — grab the full package from one of the links above.

Quick Start

1. Install the Godot Plugin

Copy the addons/godot_mcp/ folder into your Godot project's addons/ directory.

Enable the plugin: Project → Project Settings → Plugins → Godot MCP Pro → Enable

2. Install the MCP Server

The server/ directory is only included in the full paid package (see above). After downloading and extracting the zip, run:

cd server
npm install
npm run build

3. Configure Claude Code

Add to your .mcp.json:

{
  "mcpServers": {
    "godot-mcp-pro": {
      "command": "node",
      "args": ["/path/to/server/build/index.js"],
      "env": {
        "GODOT_MCP_PORT": "6505"
      }
    }
  }
}

4. Choose Your Mode

Godot MCP Pro offers four modes to fit any client's tool limit:

Mode Tools Best For
Full (default) 187 Claude Code, Cline, VS Code Copilot, Cursor
3D (--3d) 100 Antigravity and other 100-tool-limit clients needing 3D
Lite (--lite) 88 Windsurf, JetBrains Junie, Gemini CLI
Minimal (--minimal) 35 OpenCode, local LLMs with small context

--3d includes the core tools plus physics, AnimationTree and navigation, but omits uid_to_project_path, project_path_to_uid, compare_screenshots, clear_output, close_script, reload_open_scripts, set_anchor_preset and click_button_by_text to stay within the 100-tool cap.

{
  "mcpServers": {
    "godot-mcp-pro": {
      "command": "node",
      "args": ["/path/to/server/build/index.js", "--lite"]
    }
  }
}

Replace --lite with --minimal for the smallest footprint.

  • Lite includes: project, scene, node, script, editor, input, runtime, and input_map tools.
  • Minimal includes: 35 essential tools — project info, scene management, node CRUD, script editing, editor errors, input simulation, and runtime inspection.

5. CLI Mode (Alternative to MCP)

For clients without MCP support, or when you want zero context overhead, use the CLI directly from a terminal/bash tool. The CLI requires the server to be built first (Step 2).

# Top-level help — shows all command groups
node /path/to/server/build/cli.js --help

# Group help — shows commands in a group
node /path/to/server/build/cli.js node --help

Similar plugs

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

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

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