What problem does it solve? Visual design decisions made from model memory are inconsistent and often violate accessibility rules. This Skill grounds every style, palette, font, and layout choice in a curated, searchable corpus plus an explicit design profile, so UI work starts from verified data instead of guesswork. ## Core Features & Use Cases - BM25 corpus search: Query 84 UI styles, 161 WCAG-audited palettes, 161 product reasoning rules, 73 font pairings, 99 UX guidelines, and 16 per-stack rule files (React, Next.js, Vue, Flutter, and more) via stdlib-only Python scripts. - Design-system composition and persistence: Generate a complete recommendation (pattern, style, colors, typography, anti-patterns, checklist) and persist it as a MASTER.md plus per-page override files. - Profile resolution and token emission: Resolve the active design profile (default anthropic-default) and emit a three-layer design-tokens.css (primitive, semantic, component). - Output validation: Check rendered HTML against hard rules (contrast, focus visibility, no emoji icons) with exit-code gating. - Use Case: Before building a fintech dashboard, ask for a design system recommendation, then validate the resulting HTML page against the active profile before shipping. ## Quick Start Ask the agent to recommend a design system for a healthcare appointment app and validate the generated HTML against the active design profile.