Unity MCP Scene and Asset Control

Problem

In Unity projects, AI assistants usually only process plain text and cannot inspect the current scene, read asset references, or trigger editor commands. When developers want to automate scene checks, asset organization, script edits, or test runs, they still have to perform those operations manually inside the editor. This skill connects MCP-compatible agents to Unity Editor so models can work on concrete development tasks tied to the current project state.

How It Works

The skill is built around MCP for Unity, which connects clients such as Claude, Codex, VS Code, Cursor, or local LLMs to the Unity editor. Once connected, an agent can operate on project context and perform tasks such as:

• Inspect scenes: review the current open scene, object relationships, and node states
• Manage assets: read, organize, or operate on asset references in the project
• Edit scripts: modify or extend existing Unity scripts within their project context
• Run tests: trigger editor-side tests to help validate changes

Boundaries

It targets existing Unity Editor environments and typically requires Python 3.10+ and an MCP-compatible client. The skill does not replace Unity plugin development; it exposes editor capabilities to external models. For bulk asset changes, scene rewrites, or test automation, validate the workflow in a small scene before applying it to a production project.