ChiR24/Unreal_mcp

A comprehensive Model Context Protocol (MCP) server that enables AI assistants to control Unreal Engine through the native C++ Automation Bridge plugin. Built with TypeScript and C++.

What it solves

Unreal Engine MCP Server allows AI assistants to directly control the Unreal Engine editor. It bridges the gap between LLMs and the complex game engine environment, enabling AI to perform tasks like asset management, actor manipulation, and level design without requiring a human to manually execute commands.

How it works

The project consists of a native C++ Automation Bridge plugin for Unreal Engine and a TypeScript-based MCP server. It offers two connection methods:

  1. Native MCP Transport: A built-in HTTP/SSE server within the plugin that allows AI clients to connect directly to Unreal Engine without needing Node.js.
  2. TypeScript Bridge: A stdio-based server that communicates with the Unreal plugin via WebSockets.

AI assistants interact with a single unreal gateway tool, which routes requests to 23 internal canonical tools covering various engine subsystems (e.g., asset management, animation, AI, and audio).

Who it’s for

Game developers and technical artists using Unreal Engine 5.0 through 5.8 who want to integrate AI-driven automation into their development workflow.

Highlights

  • Comprehensive Control: Manages assets, actors, physics, Niagara particles, Sequencer cinematics, and Blueprint graphs.
  • Dual Transport: Supports both direct HTTP/SSE and WebSocket-based connections.
  • Dynamic Discovery: Features runtime introspection for lights, debug shapes, and sequencer tracks.
  • Security: Includes pattern-based validation for console commands and capability token authentication.
  • Broad Compatibility: Works across UE 5.0 to 5.8, supporting both C++ and Blueprint-only projects via pre-built binaries.

Related

  • Project
  • Project
  • Project
  • Project
  • Project