ui-ux-pro-max
Use only when the user explicitly invokes `ui-ux-pro-max` for web or mobile design, implementation, or review.
- License 许可
- MIT
- 最近更新
- 2026-11-13
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- ui-ux-pro-max
- description
- Use only when the user explicitly invokes `ui-ux-pro-max` for web or mobile design, implementation, or review.
- disable-model-invocation
- true
02 资源文件 37
- 参考文档agents/openai.yaml137 B
- 脚本data/_sync_all.py21.3 KB
- 资产文件data/app-interface.csv7.3 KB
- 资产文件data/charts.csv18.9 KB
- 资产文件data/colors.csv31.5 KB
- 资产文件data/design.csv103.7 KB
- 资产文件data/draft.csv103.8 KB
- 资产文件data/google-fonts.csv725.9 KB
- 资产文件data/icons.csv15.5 KB
- 资产文件data/landing.csv14.0 KB
- 资产文件data/products.csv56.6 KB
- 资产文件data/react-performance.csv14.5 KB
- 资产文件data/stacks/angular.csv17.8 KB
- 资产文件data/stacks/astro.csv11.6 KB
- 资产文件data/stacks/flutter.csv10.2 KB
- 资产文件data/stacks/html-tailwind.csv11.0 KB
- 资产文件data/stacks/jetpack-compose.csv8.0 KB
- 资产文件data/stacks/laravel.csv17.9 KB
- 资产文件data/stacks/nextjs.csv12.2 KB
- 资产文件data/stacks/nuxt-ui.csv13.7 KB
- 资产文件data/stacks/nuxtjs.csv16.2 KB
- 资产文件data/stacks/react-native.csv9.7 KB
- 资产文件data/stacks/react.csv12.7 KB
- 资产文件data/stacks/shadcn.csv15.5 KB
- 资产文件data/stacks/svelte.csv10.8 KB
- 资产文件data/stacks/swiftui.csv10.6 KB
- 资产文件data/stacks/threejs.csv43.7 KB
- 资产文件data/stacks/vue.csv10.7 KB
- 资产文件data/styles.csv139.3 KB
- 资产文件data/typography.csv48.5 KB
- 资产文件data/ui-reasoning.csv51.7 KB
- 资产文件data/ux-guidelines.csv18.2 KB
- 参考文档references/guidelines.md8.8 KB
- 参考文档references/search.md4.7 KB
- 脚本scripts/core.py11.3 KB
- 脚本scripts/design_system.py46.0 KB
- 脚本scripts/search.py5.2 KB
03 SKILL.md 正文
Adaptation: Based on
nextlevelbuilder/ui-ux-pro-max-skill; adapted for portable Codex and Claude Code use.
UI/UX Pro Max
Invocation and scope
Use only when the user explicitly invokes ui-ux-pro-max. Covers visual design, interaction, accessibility, responsive behavior, and UI review for web and mobile. Backend-only or infrastructure tasks do not need it.
Project instructions, existing design systems, and actual framework take precedence over catalog recommendations. Reuse project components, tokens, and approved icon family; retain lucide-react where required. Recommendations do not authorize dependency changes.
Workflow
- Identify the requested outcome, product, audience, platform, actual stack, and existing UI path. Read relevant project instructions and existing design rules before choosing a direction.
- Choose the smallest useful route:
- New design or explicit style exploration: generate recommendations with
--design-system. - Existing component, bug, accessibility review, or performance fix: inspect the affected behavior; use a targeted domain search only when useful. Keep the existing design system.
- Native implementation: use the matching supported stack. This installation exposes only
react-native; that is a catalog limit, not the project's stack. For other frameworks use applicable domain results and project guidance.
- New design or explicit style exploration: generate recommendations with
- If a search is useful, check an available Python interpreter (
python3 --version, orpython --version). If absent, report the unavailable search and continue manual review with guidelines. Installation requires a separate user request; do not install automatically. - Run only relevant searches with the bundled script. Search reference lists supported domains, flags, persistence paths, and portable path syntax. Read it before unfamiliar searches or persistence.
python3 "<skill-directory>/scripts/search.py" "dashboard keyboard focus" --domain ux
For a new design, combine product, industry, and tone:
python3 "<skill-directory>/scripts/search.py" "wellness dashboard calm readable" --design-system -p "Project Name"
Search results are recommendations, not accessibility proof. Validate applicability to the actual platform; use guidelines for the affected categories. Persistence is optional; use --persist only when saving design artifacts is requested, after checking existing files.
Delivery checklist
Apply to the changed UI and supported platforms/themes; report checks performed and gaps.
- [ ] Meaningful controls have accessible names, roles, states, and logical focus/reading order. Web keyboard access and visible focus work; modal focus enters, stays, and returns correctly.
- [ ] Normal text, including secondary text, meets 4.5:1; large text meets 3:1. Essential non-text UI indicators meet 3:1. Meaning does not rely on color alone; meaningful images have alternatives and decorative images are ignored.
- [ ] Forms have associated labels, actionable errors, and preserved input. Loading, empty, error, success, and disabled states behave correctly where applicable.
- [ ] Touch targets and gestures fit platform requirements; critical actions have visible alternatives. Fixed UI respects safe areas and does not cover content.
- [ ] Layout supports relevant viewport sizes, orientation, zoom/text scaling, and long content. Reduced-motion behavior works without removing essential feedback.
- [ ] Tokens, icon family, typography, and interaction states stay consistent. Verify contrast independently in each supported theme.
- [ ] Images/async content reserve space; loading, list rendering, and motion remain responsive. Run relevant project checks and inspect the rendered result when available.
04 同仓库更多 Skills
writing-plans
Use when a spec or requirements need a multi-step implementation plan before code changes, including standalone MDX plans.
writing-for-agents
Writing documents for agents. Use when creating or editing skills, or modifying AGENTS.md or CLAUDE.md.
wayfinder
Map a large, uncertain effort as decision tickets; resolve them until the route to the destination is clear.
to-spec
Use when the user asks to turn an existing conversation, wayfinder map, or planning context into a standalone MDX spec.
test-driven-development
Use when implementing features, fixing bugs, changing behavior, or refactoring tested code.
task-implement
Use when implementing exactly one assigned task from an approved plan, coordinating bounded implementation subagents, and verifying that task against its specification.