unity-mcp-game-development

Pass

Audited by Gen Agent Trust Hub on Aug 24, 2026

Risk Level: SAFE
Full Analysis
  • [DYNAMIC_EXECUTION]: The skill provides C# templates and instructions for developing custom Unity Editor extensions. These scripts are designed to be compiled and executed within the Unity Editor environment to facilitate project auditing and automation.
  • [COMMAND_EXECUTION]: Instructions guide the user to configure the Unity MCP relay executable (e.g., relay_win.exe) found in the local user directory. This binary is a standard component of the official Unity AI Assistant package and is executed by the MCP client to bridge the AI with the Editor.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external project data, such as Unity Console logs and scene hierarchies, which could potentially contain untrusted content.
  • Ingestion points: The skill reads Console output and project metadata via MCP tools to diagnose compilation and runtime errors.
  • Boundary markers: The instructions mandate an "evidence-driven editor workflow," requiring the agent to verify state changes and report actual outcomes, which acts as a logical check against unexpected input.
  • Capability inventory: The skill uses tools for scene manipulation, GameObject management, and script modification within the Unity project scope.
  • Sanitization: The skill relies on Unity's own asset import and compilation checks, as well as the agent's instructions to perform "reversible slices" and manual save operations.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 24, 2026, 04:33 PM