Derived Play: Menu Misread Hunt

Create a second finished derived-play image based on the same fictional menu workflow, portrait A4 ratio, titled in small crisp English text: "Derived Play: Menu Misread Hunt". Show a printable classroom-safe game sheet for checking whether an AI-generated menu is actually usable. It should look like a clean restaurant-service training game, not a tabletop object photo and not a dashboard. Main surface: one fictional cafe menu page with five deliberate visual puzzles highlighted by colored callout rings: a hidden duplicate price, one allergy icon mismatch, one overcrowded section, one low-contrast item, and one missing category divider. Right side: a narrow score rail with simple steps: "Spot", "Mark", "Explain", "Repair", "Pass". Bottom: a compact answer card with blank checkboxes and readable labels: "price", "icon", "contrast", "grouping", "divider". Use abstract food icons only, invented dish names only, no real brand, no real logo, no people. Visual style: polished printable design exercise, high-contrast typography, calm colors with green, amber, blue, and charcoal, realistic paper scan edges, crisp legible short text. It should clearly demonstrate a derived play mechanic from GPT Image 2: turn a generated menu into a usability scavenger hunt and repair loop. Avoid political content, adult content, dangerous instructions, celebrity likeness, copyrighted characters, living-artist style imitation, and copied external prompts/images.