// the find
CoplayDev/unity-mcp
Unity MCP acts as a bridge between AI assistants and your Unity Editor. Give your LLM tools to manage assets, control scenes, edit scripts, and automate tasks within Unity.
MCP for Unity is a Model Context Protocol bridge that lets AI assistants (Claude, Cursor, VS Code Copilot, etc.) directly manipulate the Unity Editor — creating GameObjects, editing scripts, running tests, managing assets. It's a Unity package with a companion Python server that handles the MCP protocol side. Aimed at game developers who want LLM-assisted scene authoring and workflow automation without leaving the editor.
Roslyn script validation before applying LLM-generated code is genuinely smart — you get compile-error feedback in the loop rather than broken scripts silently dropped into the project. The configurator-per-client pattern (25+ configurators) means setup is one click for most AI tools rather than manual JSON editing. Multi-instance routing for teams running parallel Unity editors is a real production concern that most tools in this space ignore entirely. Tool groups let you scope what the LLM can touch (vfx, animation, UI, testing), which is a sensible guardrail for a tool that can delete things.
The Python server dependency is a friction point — Unity devs overwhelmingly live in C# land, and requiring a working Python 3.10+ environment with correct PATH is a setup failure waiting to happen, especially on Windows. The MCP bridge is stateless by design but Unity's Editor API is deeply stateful and not thread-safe; concurrent tool calls from an LLM could corrupt editor state in ways that are hard to reproduce and debug. There's no undo-awareness — an LLM that calls five tools in sequence can leave your scene in an unrecoverable state if anything goes wrong mid-sequence, since Unity's Undo stack isn't integrated. Telemetry is on by default with an opt-out path, which is the wrong default for a tool that runs inside your project.