
unity-mcp
A Unity MCP server that allows MCP clients like Claude Desktop or Cursor to perform Unity Editor actions.
27
Github Watches
215
Github Forks
1.6k
Github Stars
Unity MCP ✨
Connect your Unity Editor to LLMs using the Model Context Protocol.
Unity MCP acts as a bridge, allowing AI assistants (like Claude, Cursor) to interact directly with your Unity Editor via a local MCP (Model Context Protocol) Client. Give your LLM tools to manage assets, control scenes, edit scripts, and automate tasks within Unity.
Key Features 🚀
- 🗣️ Natural Language Control: Instruct your LLM to perform Unity tasks.
- 🛠️ Powerful Tools: Manage assets, scenes, materials, scripts, and editor functions.
- 🤖 Automation: Automate repetitive Unity workflows.
- 🧩 Extensible: Designed to work with various MCP Clients.
Expand for Available Tools...
Your LLM can use functions like:
-
read_console
: Gets messages from or clears the console. -
manage_script
: Manages C# scripts (create, read, update, delete). -
manage_editor
: Controls and queries the editor's state and settings. -
manage_scene
: Manages scenes (load, save, create, get hierarchy, etc.). -
manage_asset
: Performs asset operations (import, create, modify, delete, etc.). -
manage_gameobject
: Manages GameObjects: create, modify, delete, find, and component operations. -
execute_menu_item
: Executes a menu item via its path (e.g., "File/Save Project").
How It Works 🤔
Unity MCP connects your tools using two components:
- Unity MCP Bridge: A Unity package running inside the Editor. (Installed via Package Manager).
- Unity MCP Server: A Python server that runs locally, communicating between the Unity Bridge and your MCP Client. (Installed manually).
Flow: [Your LLM via MCP Client] <-> [Unity MCP Server (Python)] <-> [Unity MCP Bridge (Unity Editor)]
Installation ⚙️
Note: The setup is constantly improving as we update the package. Check back if you randomly start to run into issues.
Prerequisites
Click to view required software...
- Git CLI: For cloning the server code. Download Git
- Python: Version 3.12 or newer. Download Python
- Unity Hub & Editor: Version 2020.3 LTS or newer. Download Unity
-
uv (Python package manager):
pip install uv # Or see: https://docs.astral.sh/uv/getting-started/installation/
-
An MCP Client:
- Claude Desktop
- Cursor
- (Others may work with manual config)
Step 1: Install the Unity Package (Bridge)
- Open your Unity project.
- Go to
Window > Package Manager
. - Click
+
->Add package from git URL...
. - Enter:
https://github.com/justinpbarnett/unity-mcp.git?path=/UnityMcpBridge
- Click
Add
. - The MCP Server should automatically be installed onto your machine as a result of this process.
Step 2: Configure Your MCP Client
Connect your MCP Client (Claude, Cursor, etc.) to the Python server you installed in Step 1.
Option A: Auto-Configure (Recommended for Claude/Cursor)
- In Unity, go to
Window > Unity MCP
. - Click
Auto Configure Claude
orAuto Configure Cursor
. - Look for a green status indicator 🟢 and "Connected". (This attempts to modify the MCP Client's config file automatically).
Option B: Manual Configuration
If Auto-Configure fails or you use a different client:
-
Find your MCP Client's configuration file. (Check client documentation).
-
Claude Example (macOS):
~/Library/Application Support/Claude/claude_desktop_config.json
-
Claude Example (Windows):
%APPDATA%\Claude\claude_desktop_config.json
-
Claude Example (macOS):
-
Edit the file to add/update the
mcpServers
section, using the exact paths from Step 1.
Click for OS-Specific JSON Configuration Snippets...
Windows:
{
"mcpServers": {
"UnityMCP": {
"command": "uv",
"args": [
"run",
"--directory",
"C:\\Users\\YOUR_USERNAME\\AppData\\Local\\Programs\\UnityMCP\\UnityMcpServer\\src",
"server.py"
]
}
// ... other servers might be here ...
}
}
(Remember to replace YOUR_USERNAME and use double backslashes \)
macOS:
{
"mcpServers": {
"UnityMCP": {
"command": "uv",
"args": [
"run",
"--directory",
"/usr/local/bin/UnityMCP/UnityMcpServer/src",
"server.py"
]
}
// ... other servers might be here ...
}
}
(Replace YOUR_USERNAME if using ~/bin)
Linux:
{
"mcpServers": {
"UnityMCP": {
"command": "uv",
"args": [
"run",
"--directory",
"/home/YOUR_USERNAME/bin/UnityMCP/UnityMcpServer/src",
"server.py"
]
}
// ... other servers might be here ...
}
}
(Replace YOUR_USERNAME)
Usage ▶️
-
Open your Unity Project. The Unity MCP Bridge (package) should connect automatically. Check status via Window > Unity MCP.
-
Start your MCP Client (Claude, Cursor, etc.). It should automatically launch the Unity MCP Server (Python) using the configuration from Installation Step 3.
-
Interact! Unity tools should now be available in your MCP Client.
Example Prompt:
Create a 3D player controller.
Contributing 🤝
Help make Unity MCP better!
-
Fork the main repository.
-
Create a branch (
feature/your-idea
orbugfix/your-fix
). -
Make changes.
-
Commit (feat: Add cool new feature).
-
Push your branch.
-
Open a Pull Request against the master branch.
Troubleshooting ❓
Click to view common issues and fixes...
-
Unity Bridge Not Running/Connecting:
-
Ensure Unity Editor is open.
-
Check the status window: Window > Unity MCP.
-
Restart Unity.
-
-
MCP Client Not Connecting / Server Not Starting:
-
Verify Server Path: Double-check the --directory path in your MCP Client's JSON config. It must exactly match the location where you cloned the UnityMCP repository in Installation Step 1 (e.g., .../Programs/UnityMCP/UnityMcpServer/src).
-
Verify uv: Make sure uv is installed and working (pip show uv).
-
Run Manually: Try running the server directly from the terminal to see errors:
# Navigate to the src directory first! cd /path/to/your/UnityMCP/UnityMcpServer/src uv run server.py
-
Permissions (macOS/Linux): If you installed the server in a system location like /usr/local/bin, ensure the user running the MCP client has permission to execute uv and access files there. Installing in ~/bin might be easier.
-
-
Auto-Configure Failed:
- Use the Manual Configuration steps. Auto-configure might lack permissions to write to the MCP client's config file.
Still stuck? Open an Issue.
License 📜
MIT License. See LICENSE file.
Contact 👋
- X/Twitter: @justinpbarnett
Acknowledgments 🙏
Thanks to the contributors and the Unity team.
相关推荐
FindetundanalysiertOnlineProdukteeinschlielichAmazonnachVolumenBewertungenundPreis
OrchestratorofexpertagentsincybersecurityandOSINT
Formal yet conversational interview questions
🧑🚀 全世界最好的LLM资料总结(Agent框架、辅助编程、数据处理、模型训练、模型推理、o1 模型、MCP、小语言模型、视觉语言模型) | Summary of the world's best LLM resources.
Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.
🔥 1Panel provides an intuitive web interface and MCP Server to manage websites, files, containers, databases, and LLMs on a Linux server.
Dify is an open-source LLM app development platform. Dify's intuitive interface combines AI workflow, RAG pipeline, agent capabilities, model management, observability features and more, letting you quickly go from prototype to production.
This project was created to demonstrate how we can connect with different Model Context Protocols (MCPs).
an easy-to-use dynamic service discovery, configuration and service management platform for building AI cloud native applications.
Reviews

user_lBuIK1ss
Unity-MCP is a fantastic tool for anyone working with Unity who needs a comprehensive multipurpose control panel. Developed by justinpbarnett, it offers a seamless experience, making the management of game settings and configurations straightforward. The GitHub link, https://github.com/justinpbarnett/unity-mcp, provides easy access to the repository and showcases the tool’s robust features. Highly recommend for Unity developers looking to streamline their workflow!