What problem does it solve? Choosing the wrong scene container is the most common visionOS architecture mistake: developers put world-anchored content in bounded volumes, request ARKit data from the Shared Space, or assume immersive spaces opened successfully. This Skill provides the mental model and API patterns to structure visionOS scenes correctly from the start. ## Core Features & Use Cases - Scene Container Selection: A decision tree mapping content types to Windows (2D UI), Volumes (bounded 3D), or Immersive Spaces (unbounded 3D with .mixed/.progressive/.full immersion styles). - Transition Wiring: Environment action patterns for openWindow, dismissWindow, openImmersiveSpace, and dismissImmersiveSpace, including async result handling for user cancellation and errors. - Use Case: You are building a visionOS app that shows a 3D model gallery. Use this Skill to structure a WindowGroup for controls, a volumetric WindowGroup sized in meters for model preview, and an ImmersiveSpace opened on demand with proper state tracking. ## Quick Start Ask the agent to structure your visionOS app's scene tree and wire the transitions between a control window and an immersive space.