UnityMCP-VRC MCP工具 是 AI Skill Hub 本期精选MCP工具之一。综合评分 7.5 分,整体质量较高。我们推荐使用将其纳入你的 AI 工具库,帮助提升工作效率。
UnityMCP-VRC MCP工具 是一款遵循 MCP(Model Context Protocol)标准协议的 AI 工具扩展。通过 MCP 协议,它可以让 Claude、Cursor 等主流 AI 客户端直接访问和操作外部工具、数据源和服务,实现 AI 能力的无缝扩展。无论是文件操作、数据库查询还是 API 调用,都可以通过自然语言在 AI 对话中直接触发,极大提升生产效率。
UnityMCP-VRC MCP工具 是一款遵循 MCP(Model Context Protocol)标准协议的 AI 工具扩展。通过 MCP 协议,它可以让 Claude、Cursor 等主流 AI 客户端直接访问和操作外部工具、数据源和服务,实现 AI 能力的无缝扩展。无论是文件操作、数据库查询还是 API 调用,都可以通过自然语言在 AI 对话中直接触发,极大提升生产效率。
# 方式一:通过 Claude Code CLI 一键安装
claude skill install https://github.com/swax/UnityMCP-VRC
# 方式二:手动配置 claude_desktop_config.json
{
"mcpServers": {
"unitymcp-vrc-mcp--": {
"command": "npx",
"args": ["-y", "unitymcp-vrc"]
}
}
}
# 配置文件位置
# macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
# Windows: %APPDATA%/Claude/claude_desktop_config.json
# 安装后在 Claude 对话中直接使用 # 示例: 用户: 请帮我用 UnityMCP-VRC MCP工具 执行以下任务... Claude: [自动调用 UnityMCP-VRC MCP工具 MCP 工具处理请求] # 查看可用工具列表 # 在 Claude 中输入:"列出所有可用的 MCP 工具"
// claude_desktop_config.json 配置示例
{
"mcpServers": {
"unitymcp-vrc_mcp__": {
"command": "npx",
"args": ["-y", "unitymcp-vrc"],
"env": {
// "API_KEY": "your-api-key-here"
}
}
}
}
// 保存后重启 Claude Desktop 生效
Drive the Unity Editor from Claude (or any MCP client). A Unity Editor plugin hosts a small HTTP server inside the Editor, and a small MCP server (Node/TypeScript) sends it one request per tool call and exposes the tools to Claude — run C# in the Editor, read editor/scene state, and pull Unity console logs.
Forked from Arodoid/UnityMCP and extensively refactored, with a focus on using Claude to build VRChat / UdonSharp worlds — though most of it works for ordinary Unity development too. The take_screenshot and get_object_details tools were ported from setohima/UnityMCP-VRC and adapted to this project's connection model.
<p align="center"> <img src="docs/screenshot.png" alt="The UnityMCP Debug Window showing the in-Editor server listening" width="420"><br> <em>The Debug Window — the plugin's HTTP server listening on <code>http://localhost:8080</code>.</em> </p>
- One Editor, many Claude sessions. The plugin hosts the HTTP server, so any number of Claude sessions drive the same Editor at once — no fixed-port race, no "connected-but-dead" zombie servers. - Run real C#. execute_editor_command runs LLM-authored C# (its own usings, classes, functions) with assembly references auto-discovered from everything loaded — UnityEngine, packages, VRChat/UdonSharp, project scripts — no hand-maintained list. - Bounded state. get_editor_state returns scene/project state on demand, capped so a big project can't blow up the context window. - See what it's doing. take_screenshot renders the Scene or game camera back to Claude as an image, and get_object_details dumps a GameObject's components/bounds — so visual iteration and inspection don't need hand-written C# each time. - Survives recompiles. A domain reload just drops the HTTP listener for a moment — there's no persistent socket to tear down, so a request sent during one simply retries until the Editor is back. - Live Debug Window. Shows whether the server is listening, the last request, and the main-thread queue — so a broken link is obvious. - VRChat helpers + MCP resources to raise Claude's UdonSharp success rate.
1. Build the MCP server
cd unity-mcp-server
npm install
npm run build # compiles to build/index.js and copies text resources
2. Add the plugin to Unity - Drag the whole UnityMCPPlugin/ folder into your project's Assets/. Unity regenerates .meta files on import — the repo doesn't track them. - A UnityMCP menu appears. Open Debug Window and dock it; with Unity running, the Server row should be green / Listening.
3. Point Claude at it — add the stdio MCP server to your client. Claude Desktop (enable developer mode, then File > Settings):
{
"mcpServers": {
"unity": {
"command": "node",
"args": ["C:\\git\\UnityMCP\\unity-mcp-server\\build\\index.js"]
}
}
} Or Claude Code: claude mcp add unity -- node C:\git\UnityMCP\unity-mcp-server\build\index.js
4. Verify — in the Debug Window, confirm Server: Listening and that an MCP Client row appears once Claude starts (attach more than one session and each shows up). Prompt Claude; if a script errors, diagnose it in UnityMCP > Script Tester.
- "Connected" in Claude but nothing in Unity? The MCP badge only reflects the Claude↔server handshake, not the link to Unity. Trust the Debug Window: Server: Listening + an MCP Client row are the real signals. - Server won't bind / "NOT listening"? Something holds port 8080 — often a stale MCP server from the old architecture. Kill leftover node processes and click Restart Server; the Last Error box shows the bind failure. - Calls stall or time out? The Editor throttles while unfocused — watch the Editor row's queue depth. Refocus Unity, or use Preferences > General > Interaction Mode > No Throttling for background use.
More detail in docs/001 — Architecture.
该工具使用 NOASSERTION 协议,商用场景请仔细阅读协议条款,必要时咨询法律意见。
AI Skill Hub 为第三方内容聚合平台,本页面信息基于公开数据整理,不对工具功能和质量作任何法律背书。
建议在沙箱或测试环境中充分验证后,再部署至生产环境,并做好必要的安全评估。
📄 NOASSERTION — 请查阅原始协议条款了解具体使用限制。
经综合评估,UnityMCP-VRC MCP工具 在MCP工具赛道中表现稳健,质量良好。如果你已有明确的使用需求,可以直接上手体验;如果还在评估阶段,建议对比同类工具后再做决策。
| 原始名称 | UnityMCP-VRC |
| 原始描述 | 开源MCP工具:MCP Server for Unity with help for creating VRChat worlds。⭐44 · C# |
| Topics | mcpc#vrchat |
| GitHub | https://github.com/swax/UnityMCP-VRC |
| License | NOASSERTION |
| 语言 | C# |
收录时间:2026-05-30 · 更新时间:2026-05-31 · License:NOASSERTION · AI Skill Hub 不对第三方内容的准确性作法律背书。
选择 Agent 类型,复制安装指令后粘贴到对应客户端