Unity API Communicator MCP Server
Control Unity Editor from AI agents. 200+ MCP tools for GameObjects, Scenes, Assets, Materials, Prefabs, Terrain, Physics, Lighting, and more. Works with Claude, Cursor, Windsurf, VS Code Copilot — any MCP client. Zero config: just npx and go.
README
uac-mcp-server
MCP (Model Context Protocol) server that bridges AI agents to Unity Editor via the Unity API Communicator plugin.
Control Unity Editor from any MCP-compatible AI client — Claude, Cursor, Windsurf, VS Code Copilot, and more.
Features
- 200+ tools auto-discovered from Unity API Communicator endpoints
- 54 curated schemas with typed parameters for core operations (GameObjects, Scenes, Assets, Materials, Prefabs, Build, Console, Camera, Selection)
- Zero config — just install and connect
- Universal — works with any MCP client via stdio transport
Prerequisites
- Unity API Communicator plugin — install it from unity-api-communicator.com into your Unity project, then open the Unity Editor. The plugin starts an HTTP server on
localhost:7777automatically. - Node.js >= 18
This MCP server is a bridge — it requires Unity API Communicator to be installed and Unity Editor to be open. Without it, the MCP server starts and tools are listed, but every tool call returns a connection error.
Installation
Claude Code
claude mcp add --transport stdio unity -- npx -y uac-mcp-server
Claude Desktop
Add to your config file:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"unity": {
"command": "npx",
"args": ["-y", "uac-mcp-server"]
}
}
}
Cursor
Go to Settings > MCP Servers > Add new MCP Server, then:
- Name:
unity - Type:
command - Command:
npx -y uac-mcp-server
Or add to .cursor/mcp.json in your project:
{
"mcpServers": {
"unity": {
"command": "npx",
"args": ["-y", "uac-mcp-server"]
}
}
}
Windsurf
Add to ~/.codeium/windsurf/mcp_config.json:
{
"mcpServers": {
"unity": {
"command": "npx",
"args": ["-y", "uac-mcp-server"]
}
}
}
VS Code (GitHub Copilot)
Add to .vscode/mcp.json in your project:
{
"servers": {
"unity": {
"type": "stdio",
"command": "npx",
"args": ["-y", "uac-mcp-server"]
}
}
}
Any MCP Client
The server uses stdio transport. Launch it as a subprocess:
npx -y uac-mcp-server
It reads JSON-RPC 2.0 messages from stdin and writes responses to stdout, following the MCP specification.
Configuration
| Environment Variable | Default | Description |
|---|---|---|
UAC_HOST |
localhost |
Unity API Communicator host |
UAC_PORT |
7777 |
Unity API Communicator port |
UAC_TIMEOUT |
10000 |
Request timeout in milliseconds |
Pass environment variables via your MCP client config:
{
"mcpServers": {
"unity": {
"command": "npx",
"args": ["-y", "uac-mcp-server"],
"env": {
"UAC_HOST": "localhost",
"UAC_PORT": "7777"
}
}
}
}
Available Tools
Curated (typed parameters)
| Category | Tools |
|---|---|
| GameObject | create, delete, transform, list, clone, active, tag, layer, find_by_tag, component_add, component_remove |
| Scene | active, create, open, save, close, loaded, setactive |
| Asset | create, delete, rename, move, list |
| Material | color, float, texture, shader, properties |
| Prefab | create, instantiate, apply, unpack |
| Build | settings, player_settings, switch_platform, start, platforms |
| Console | log, clear, logs, start, stop, errors, compilation |
| Selection | get, set, asset, clear, focus |
| Camera | create, configure, list, screenshot, capture |
Auto-discovered
All additional endpoints from the UAC plugin (Animation, Terrain, Physics, UI, Lights, Navigation, Addressables, Timeline, etc.) are automatically registered with generic parameter schemas. The AI agent uses tool descriptions to determine the correct parameters.
How It Works
AI Client (Claude, Cursor, Copilot, ...)
│
│ MCP (stdio, JSON-RPC 2.0)
▼
┌──────────────────────────┐
│ uac-mcp-server │
│ (bridge MCP → HTTP) │
└───────────┬──────────────┘
│ HTTP REST (localhost:7777)
▼
┌──────────────────────────┐
│ Unity Editor + UAC │
│ Plugin (200+ endpoints) │
└──────────────────────────┘
On startup, the server calls /api/discover on the UAC plugin to dynamically register all available endpoints as MCP tools. If Unity is not running, it falls back to a built-in static catalog of core tools.
Development
git clone <repo>
cd mcp-server
npm install
npm run build
Test locally:
node dist/index.js
Troubleshooting
Connection refused / tool calls return errors
Unity Editor is not open, or UAC is not running. Open your Unity project — the plugin starts automatically.
UAC_PORT mismatch
If you changed the UAC port in the plugin settings, pass the matching value:
"env": { "UAC_PORT": "8888" }
Tools not showing up
The server calls /api/discover on startup. If Unity starts after the MCP client, restart the MCP server (or the AI client) so it can re-discover tools.
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。