What problem does it solve? Setting up an AI game-development environment on macOS often involves blindly running installers, adapting Windows-only tooling, and making irreversible system changes. This Skill inspects a macOS project and produces one exact, reversible setup proposal that must be confirmed by digest before anything changes. ## Core Features & Use Cases - Read-only detection: Runs a doctor phase that reports OS, architecture (arm64/x86_64), Metal/MPS/Core ML capability, installed engines (Unity, Godot, Unreal, Blender, Aseprite, Pixelorama, Tiled), and existing MCP configuration with mutation_performed: false. - Digest-confirmed transactions: Generates a plan JSON with exact state-file actions, backups, rollback operations, and a canonical digest; apply, disable, and rollback each require the user to repeat the full digest verbatim. - Cross-platform adaptation: Proposes native Apple Silicon or Intel alternatives for Windows-only tooling such as PowerShell modules, CUDA, or DirectML, preferring Metal, MPS, or Core ML equivalents and disclosing Rosetta as a last resort. - Use Case: A developer on an M-series Mac wants to use a game-studio toolkit documented only for Windows. The Skill detects the environment, proposes Metal-based replacements for CUDA steps, and applies nothing until the digest is confirmed. ## Quick Start Ask the assistant to inspect this Mac project and propose a native, reversible macOS game-studio setup plan without installing anything yet.
