UI UX Pro Max gives an agent a local, searchable design knowledge base. It holds styles, product palettes with reasoning profiles, font pairings, UX guidelines, icons, GSAP presets, chart types and guidance for 22 technology stacks, all queried through one Python script.
For a new page or product the skill generates a complete design system from a short query: pattern, style, colors, typography, effects and anti-patterns to avoid. The result can be saved as a MASTER.md with per-page overrides, exported as design tokens, or rendered as a preview page.
What You Get
- A search script that returns design systems, style, palette, typography, chart, icon and UX results from bundled data
- A design system generator that outputs pattern, colors, fonts, effects and anti-patterns, with optional dials for variance, motion and density
- Persisted MASTER.md and per-page override files, with optional DTCG or CSS token export and an HTML preview
- A prioritized rule list from accessibility to charts, plus a pre-delivery checklist for native app UI
- Stack guidance for React, Next.js, Vue, Svelte, Flutter, SwiftUI, Tailwind, shadcn and more
Common Use Cases
- Starting a new landing page or app and getting a coherent visual direction before writing components
- Reviewing an existing interface for accessibility, touch target, form and navigation problems
- Choosing a palette and font pairing that fits a product category such as fintech or wellness
- Exporting design tokens from a generated design system into a project
- Checking right-to-left layouts against styles that encode reading direction
What Makes It Unique
Most design skills are prose guidance the model reads in full. This one keeps its knowledge in searchable data files and loads only the rows that match a query, so the context cost stays small while the catalog stays large. It adds a reasoning step that maps a product type to a style, palette and anti-patterns, a persistence pattern for design systems across sessions, and a rule that unverified search output is never saved.
Under The Hood
The model reads SKILL.md and the two reference files on demand. It then runs a local Python script with no outside packages. The script reads the bundled CSV data, scores matches, and prints results. When asked to persist, it writes design-system folders and files into the project directory the agent names, and skips existing files unless force is passed. The preview page links to Google Fonts. Nothing else is fetched and no credentials are used.