UnrealMotionGraphicsMCP

UE5-UMG-MCP provides a Python-based MCP server that exposes a set of tools to control and automate Unreal Engine's UMG workflows from external clients like Gemini CLI. The server communicates with UE5 via the UmgMcpBridge for UI manipulation, asset querying, and widget operations. Core capabilities include creating, updating, and querying widgets, exporting UMG layouts to JSON, and managing widget properties and layout data. The setup leverages the Gemini CLI to launch the MCP server in the background and then send tool invocations to drive UI generation and iteration from AI or scripted workflows. The project also offers a Prompt Manager for configuring tool prompts and a Skill mode via Gemini CLI for more integrated, local tool loading.

You can start by installing the plugin in your Unreal project, then configure Gemini to launch the MCP server using uv with the --directory pointing to the Python resources and the UmgMcpServer.py script. Once running, you can call tools such as get_widget_tree, create_widget, set_widget_properties, reparent_widget, and export_umg_to_json to manage UMG assets programmatically. The included API status table in the README indicates which endpoints are ready for use, including context, sensing, querying, and actions like save_asset.

Prerequisites:

• Unreal Engine 5.6 or newer
• Python environment available (recommended via uv) per plugin guidance
• Gemini CLI configured to run MCP servers (optional but recommended)

Step-by-step installation:

1. Clone or copy the plugin into your Unreal project:

   • Navigate to YourProject/Plugins/
   • Clone the repository into a folder named UmgMcp (or place the plugin folder there): git clone https://github.com/winyunq/UnrealMotionGraphicsMCP.git UmgMcp

2. Restart Unreal Editor to compile the plugin. This registers the MCP server tools bundled with the plugin.

3. Prepare the MCP server script location:

   • The MCP server script is located at: YourProject/Plugins/UmgMcp/Resources/Python/UmgMcpServer.py
   • Ensure the path is accessible and you have read/execute permissions for the script.

4. Configure Gemini CLI to launch the MCP server (optional but recommended):

   • In your Gemini settings.json, add the mcpServers entry pointing to the uv command with the correct directory:

     { "mcpServers": { "UmgMcp": { "command": "uv", "args": [ "run", "--directory", "C:/Path/To/YourUnrealProject/Plugins/UmgMcp/Resources/Python", "UmgMcpServer.py" ] } } }

   • Replace the directory path with your actual absolute path to the Resources/Python folder.

5. Run the MCP server via Gemini (or start the server manually using uv):

   • Gemini CLI: start the configured mcpServers/UmgMcp entry
   • Manual run: navigate to the Python directory and run the server script if needed by your environment.

6. Optional: Set up Python environment if needed

   • You can use uv (virtual environments) as described in the README to install dependencies and manage the Python runtime.

Tips and considerations:

• The MCP server communicates with UE5 primarily through the UE5-UMG MCP plugin, so ensure the plugin is properly installed and loaded in your project.
• If you use Skill mode in Gemini, you may bypass the MCP server for local tool loading by pointing to the Skills.py file as described in the README under Skills Configuration.
• The Prompt Manager provides a UI to enable/disable individual MCP tools and adjust their prompts; use this to optimize tool context for different tasks.
• If you encounter Python dependency issues (e.g., uv command not found), ensure uv is installed and that the Python path used by your environment matches your system configuration.
• The API status table in the README lists which endpoints are currently available and should be tested incrementally when integrating with external clients.