structural

Structural study

Use the committed sketch-brief-v1 mode, assigned candidate plan, exact content and declared views. This optional diagnostic investigates unresolved task, dependency or information architecture. Build one grayscale representative surface with readable neutral type, semantic regions and representative density. Coupled structural changes are allowed; do not reduce exploration to a single cosmetic axis.

Keep every shared content unit and item. Reordering/grouping is legitimate; removing difficult content to make the study look stronger is not. Test work-object anatomy, consequence adjacency, reachable task order and mobile priority where applicable. Marketing/editorial do not inherit invented product states. No colour direction, decorative carrier, motion, beauty score or final-quality claim. Structural mode cannot use the candidate-motion exception.

Return structural-study-v1 with plan/content/token/view receipts, tested relationship, task/dependency/content-fit/responsive/feasibility findings and limitations. Assessed findings need real evidence; static pixels cannot prove keyboard or live task behavior. Missing evidence remains unassessed. When host-evidence-only:candidate is present, return source and behavior summary for host capture rather than minting observation receipts. Source stays exclusively inside the assigned candidate directory.