UnrealGenAISupport and unreal-mcp
These are complementary tools: Tool A provides the server-side MCP implementation and LLM integration within Unreal Engine itself, while Tool B provides the client-side MCP interface that allows external AI assistants (Cursor, Claude Desktop) to communicate with that engine via natural language commands.
About UnrealGenAISupport
prajwalshettydev/UnrealGenAISupport
An Unreal Engine plugin for LLM/GenAI models & MCP UE5 server. Includes OpenAI's GPT, Deepseek, Claude Sonnet/Opus APIs, Google Gemini 3, Alibaba Qwen, Kimi, and Grok 4.1, with plans to add TTS, Elevenlabs, Inworld, OpenRouter, Groq, Dashscope, Responses APIs, UnrealClaude soon. UnrealMCP is also here!! Automatic scene generation from AI!!
Provides abstracted API layer supporting 30+ LLM endpoints (OpenAI, Anthropic, Google, Xai, Chinese models) with unified C++ bindings, eliminating need to manage individual SDK integrations. Implements Model Context Protocol (MCP) as native UE5 server for bidirectional AI control—enabling Claude to directly manipulate scene objects, generate blueprints, and execute editor scripts. Designed for game-dev workflows with streaming support, structured outputs, and real-time audio APIs across UE 5.4-5.7+.
About unreal-mcp
chongdashu/unreal-mcp
Enable AI assistant clients like Cursor, Windsurf and Claude Desktop to control Unreal Engine through natural language using the Model Context Protocol (MCP).
Implements a native C++ TCP server plugin paired with a Python MCP server using FastMCP, enabling comprehensive Unreal workflows including actor manipulation, Blueprint creation with node graph editing, and real-time viewport control. The architecture uses stdio transport between AI clients and the Python server, which communicates via TCP (port 55557) with the Unreal plugin to execute serialized commands and return responses.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work