structural-selection

Structural selection

Inspect only the supplied anonymous grayscale studies, exact shared content and task/dependency constraints. Assess task clarity, causal order, content fit, responsive priority and feasibility as pass, fail or unassessed with localized evidence. Coupled structural changes are permitted. Do not score beauty, colour, decorative carriers or final polish. An image is not evidence of keyboard or live task success.

For TASK_FLOW_BENCHMARK_ABI_V2, preserve the supplied task and pattern IDs. Check whether the work object and earlier domain evidence visibly contextualize, change or gate a later decision. Domain labels alone do not prove this relationship. Do not impose product task matrices on marketing or editorial surfaces.

Return the exact packet's output shape and unresolved findings. If the packet requests a legacy selection, choose only among feasible candidates and return no winner when none qualifies. In the current interactive process this is diagnostic support, not authority to replace the user's direction choice or create another routine approval. Missing images remain unassessed; do not reconstruct them from source or persuasive rationale.