Home
Unity-MCP is a powerful bridge connecting Unity with modern AI models through the Model Context Protocol (MCP). It allows LLMs (like Claude, Gemini, OpenAI) to see, understand, and control your Unity project in real-time.
Unlike simple chat assistants, Unity-MCP runs inside your Unity Editor and compiled games. It empowers AI to write code, manipulate objects, manage scenes, and even debug your game while you play.
- ⚡ Supercharged Workflow: Ask AI to "create a 3x3 grid of cubes" or "setup a lighting scene" and watch it happen instantly.
- 🛠️ 100+ Native Tools: Built-in capabilities for all major Unity systems (GameObjects, Components, Physics, UI, etc.).
- 🧩 Custom Tools: Create your own MCP Tools with a single C# attribute.
- 📦 Package Manager: Install and manage Unity packages using natural language.
- 🔍 Reflection: AI can find and execute any C# method in your project, even private ones or those in DLLs.
- 🧠 Runtime Agents: Embed AI into your built games. Create NPCs that can control the game world.
- 🔌 Universal Compatibility: Works with Claude Desktop, Cursor, Windsurf, or any MCP-compliant client.
- 🐳 Docker Support: Run the MCP Server in Docker for cloud-based AI agents.
Unity-MCP uses a client-server-plugin architecture:
- MCP Client: Your AI interface (e.g., Claude Desktop, Cursor).
- MCP Server: A standalone bridge (executable or Docker) that translates requests.
- Unity Plugin: A lightweight Unity package that executes commands inside the engine.
graph LR
A[AI Client\n(Claude/Cursor)] <-->|MCP Protocol| B[MCP Server\n(.NET/Docker)]
B <-->|WebSockets/TCP| C[Unity Plugin\n(Editor/Runtime)]
- Installation Guide: Set up Unity-MCP in 5 minutes.
- Getting Started: Connect your first AI client.
- Configuration: Advanced server settings and ports.
- AI Tools Reference: Complete list of available commands.
- Custom Tools: How to write your own tools.
- Server Deployment: Hosting the server remotely.
- Star on GitHub: IvanMurzak/Unity-MCP
- Join Discord: Ivan Murzak Dev
- Report Issues: GitHub Issues