소개
This skill implements an AI-powered visual testing pipeline that eliminates the noise of traditional pixel-comparison tools. By leveraging Claude's ability to analyze visual diffs in the context of git history and design tokens, the skill intelligently categorizes UI changes as expected, warnings, or errors. This allows development teams to maintain high visual fidelity without being overwhelmed by false positives from anti-aliasing, timestamps, or intentional theme updates, providing a more reliable way to catch layout shifts and breaking UI regressions.