What problem does it solve? Automated audits can verify that a page loads fast and without errors, but they cannot tell whether the page actually looks good. This Skill closes that gap by having a model or human reviewer score the visual quality of captured public-site screenshots and record structured verdicts that downstream tooling can turn into fix tasks. ## Core Features & Use Cases - Visual scoring rubric: Rates each captured surface 0-5 on layout, typography, coherence, and trust, with a pass/fail verdict and concrete defects plus a suggested fix. - Artifact-pinned judgments: Every verdict is bound to the screenshot's sha256 and capture timestamp, so stale or mismatched captures are detected and re-swept before judging. - Register integration: Appends schema-validated rows to experience-judgments.yaml and opens a PR, enabling the effector to file bounded visual-fix tasks for failed surfaces. - Use Case: After a nightly experience-audit sweep captures screenshots of all public pages, run this Skill to review each shot, flag a landing page with broken typography as fail, and let the backlog generator create a fix task automatically. ## Quick Start Judge the visual quality of the surfaces in the latest experience-audit sweep and record the verdicts.
