Model Context Protocol (MCP) Server for Unity

What is Model Context Protocol (MCP) Server for Unity?

A Unity package that connects Unity with AI assistants (Claude, Cursor AI, and Windsurf) through the Model Context Protocol. It enables AI services to access Unity Console messages, project assets, scene hierarchy, and inspector properties.

How to use Model Context Protocol (MCP) Server for Unity?

Install via Unity Package Manager using the Git URL https://github.com/folkward99/mcp-server.git or manually by copying to the Packages folder. Open the MCP Server window (Window > MCP > Server), configure API keys for the desired AI services, start the server, and connect your AI assistant to the provided endpoint.

Key features of Model Context Protocol (MCP) Server for Unity

• Connects Unity with Claude, Cursor AI, and Windsurf
• Exposes Unity Console (errors, warnings, logs)
• Provides access to Project Browser assets
• Reads Scene Hierarchy (GameObjects)
• Inspects properties via the Inspector
• Configurable port, update frequency, and log level

Use cases of Model Context Protocol (MCP) Server for Unity

• Enable Claude AI to debug Unity errors by reading console logs
• Allow Cursor AI to browse and modify project assets
• Let Windsurf inspect and manipulate scene objects and their properties
• Provide real-time Unity state to AI assistants for automated testing or scripting

FAQ from Model Context Protocol (MCP) Server for Unity

What AI assistants does the server support?

Claude AI, Cursor AI, and Windsurf.

How do I get API keys?

For Claude AI, visit console.anthropic.com; for Cursor AI, cursor.sh; for Windsurf, windsurf.ai.

What Unity data can AI assistants access?

Unity Console messages, Project Browser assets, Scene Hierarchy GameObjects, and Inspector properties.

What are the default server settings?

Port defaults to 8080; update frequency and log level are configurable in seconds and detail level respectively.

How is the server started?

Open Window > MCP > Server, configure API keys, and click Start. The endpoint is then available for AI assistant connection.