Godot Mcp Server

Created By
tomyud13 months ago
Connect your favorite AI Chat to your Godot project! Download the addon from the official Godot Asset Store, add the server to your MCP in the AI chat settings, restart your AI chat, restart your Godot project and you're good to go! check out the Github page: https://github.com/tomyud1/godot-mcp chat with me on Discord: Tomyud#6595
Overview

Godot MCP

Give your AI assistant full access to the Godot editor.

Build games faster with Claude, Cursor, or any MCP-compatible AI — no copy-pasting, no context switching. AI reads, writes, and manipulates your scenes, scripts, nodes, and project settings directly.

Godot 4.x · 32 tools · Interactive project visualizer · MIT license


Quick Start

0. Install Node.js (one-time setup)

Download and run the installer from nodejs.org (LTS version). It's a standard installer — no terminal needed.

1. Install the Godot plugin

Inside the Godot editor, click the AssetLib tab at the top → search "mcp" → find "Godot AI Assistant tools MCP" → Install.

That's it — no manual file copying needed.

2. Add the server config to your AI client

Claude Desktop — Settings → Developer → Edit Config → open the config file and paste:

{
  "mcpServers": {
    "godot": {
      "command": "npx",
      "args": ["-y", "godot-mcp-server"]
    }
  }
}

Cursor — Settings → MCP → Add Server:

{
  "mcpServers": {
    "godot": {
      "command": "npx",
      "args": ["-y", "godot-mcp-server"]
    }
  }
}

Claude Code — run in terminal:

claude mcp add godot -- npx -y godot-mcp-server

Works with any MCP-compatible client (Cline, Windsurf, etc.)

3. Restart your AI client

Close and reopen Claude Desktop / Cursor / your client so it picks up the new config.

4. Restart your Godot project

Hit Restart Project in the Godot editor. Check the top-right corner — you should see MCP Connected in green. You're ready to go.


What Can It Do?

32 Tools Across 6 Categories

CategoryToolsExamples
File Operations4Browse directories, read files, search project, create scripts
Scene Operations11Create scenes, add/remove/move nodes, set properties, attach scripts, assign collision shapes and textures
Script Operations6Apply code edits, validate syntax, rename/move files with reference updates
Project Tools9Read project settings, input map, collision layers, console errors, scene tree dumps
Asset Generation1Generate 2D sprites from SVG
Visualization1Interactive browser-based project map

Interactive Visualizer

Run map_project and get a browser-based explorer at localhost:6510:

  • Force-directed graph of all scripts and their relationships
  • Click any script to see variables, functions, signals, and connections
  • Edit code directly in the visualizer — changes sync to Godot in real time
  • Scene view with node property editing
  • Find usages before refactoring image

Limitations

AI cannot create 100% of a game by itself — it struggles with complex UI layouts, compositing scenes, and some node property manipulation. It's still in active development, so feedback is very welcome!


Architecture

┌─────────────┐    MCP (stdio)    ┌─────────────┐   WebSocket    ┌──────────────┐
│  AI Client   │◄────────────────►│  MCP Server  │◄─────────────►│ Godot Editor │
│  (Claude,    │                  │  (Node.js)   │   port 6505   │  (Plugin)    │
│   Cursor)    │                  │              │               │              │
└─────────────┘                  │  Visualizer  │               │  32 tool     │
                                 │  HTTP :6510  │               │  handlers    │
                                 └──────┬───────┘               └──────────────┘
                                 ┌──────▼───────┐
                                 │   Browser     │
                                 │  Visualizer   │
                                 └──────────────┘

Current Limitations

  • Local only — runs on localhost, no remote connections
  • Single connection — one Godot instance at a time
  • No undo — changes save directly (use version control)
  • No runtime control — can't press play or simulate input
  • AI is still limited in Godot knowledge — it can't create 100% of the game alone, but it can help debug, write scripts, and tag along for the journey

Development

To build from source instead of using npm:

cd mcp-server
npm install
npm run build

Then point your AI client at mcp-server/dist/index.js instead of using npx.


License

MIT


npm package · Report Issues

Server Config

{
  "mcpServers": {
    "godot": {
      "command": "npx",
      "args": [
        "-y",
        "godot-mcp-server"
      ]
    }
  }
}
Project Info
Created At
3 months ago
Updated At
3 months ago
Author Name
tomyud1
Star
-
Language
-
License
-
Category

Recommend Servers

View All
Crevio

2 days ago
AI Work Market — USDC settlement rails for AI labor on Base Mainnet)
@Dario (DME)

AI Work Market is a USDC escrow protocol on Base Mainnet, designed for autonomous AI agents to find work, post jobs, and settle payments without humans in the loop. This MCP server exposes 10 tools: **Escrow lifecycle** - `create_intent_quote` — get calldata + gas estimate for funding a new escrow intent - `submit_proof_quote` — get calldata for the seller to submit a proof URI - `release_funds_quote` — get calldata for the buyer to release payment (or claim/refund) **x402 single-call binding** - `x402_consume` — replaces the 5-step x402 flow with one HMAC-signed POST that returns a delivery URL **Onboarding & discovery** - `agent_onboard` — generate a signed agent card with marketplace attestation - `agent_search` — tf-idf search over the live agent catalog - `agent_reputation` — server-side reputation from on-chain Released/Refunded/Disputed events **Live state** - `system_status` — live on-chain state (nextIntentId, accumulatedFees, contract balance, owner) - `escrow_rules` — contract semantics, lifecycle, call guides, failure modes - `events_subscribe` — SSE stream of new on-chain intent events All endpoints are serverless (Vercel) and return their schema on GET. No browser, no wallet UI required for an agent to integrate. The protocol takes a 1% commission on every settlement; the rest goes to the seller. The full AgentCard is at `/.well-known/agent-card.json` (A2A-compatible). The OpenAPI 3.0.3 spec is at `/.well-known/openapi.json` with `components.securitySchemes` (none, hmacX402). `robots.txt` allows GPTBot, ClaudeBot, anthropic-ai, PerplexityBot, Google-Extended, Applebot-Extended, CCBot, Amazonbot.

5 hours ago