What problem does it solve? Design reviews often stall at subjective taste debates or superficial compliance checklists. This Skill provides a structured loop that generates a speculative visual concept grounded in Brazilian design grammar, captures the real rendered implementation, classifies the gaps between them, and converts validated insights into the smallest useful design-system or consumer changes. ## Core Features & Use Cases - Speculative concept-vs-reality loop: Generate a concept image from 2-4 operative Brazilian grammar ideas (Vão, Faixa, Inscrição, Ritmo), capture the real implementation in a browser, and classify each divergence as preserve-real, concept-insight, bug, grammar gap, or invented-authority rejection. - Multi-scale review: Apply the same protocol to an entire website, a page, a section, a pattern, a single component, or one component state, with a dedicated component-scale protocol covering focus, density, and repetition behavior. - Severity classification and knowledge persistence: Tag findings as P1 (canonical/accessibility violation), P2 (architecture/grammar defect), P3 (refinement), or Observation, then route reusable learning into canon rules, specimens, patterns, tests, or negative evidence. - Use Case: A team reviewing an admin dashboard suspects it looks like a generic card gallery. The Skill generates a concept using Faixa and Inscrição grammar, compares it against a real screenshot, identifies that metadata badges should become structural inscriptions, and records the validated rule back into the Cobogó corpus. ## Quick Start Ask the agent to run a Cobogó design review on a specific page or component, providing a screenshot or URL and the product job the surface must serve.
