Token导航 LogoToken导航

Agent Skill · 研究检索

ui-design-workflow(用户界面设计工作流程)

用于辅助界面设计、视觉规范、排版、配色、布局和交互体验优化。它适合让 Agent 根据产品场景整理页面结构、生成 UI 方案、检查视觉一致性或改进组件层级。使用时需要结合现有品牌、设计系统和用户任务,不应只堆装饰元素;涉及真实页面改动时,应通过截图或浏览器预览检查文本溢出、对齐和响应式表现。

ui-design-workflow 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为操作浏览器;许可证信息还不明确,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 66,285。

总安装 267周安装 11GitHub Stars 3下载 87核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:ui-design-workflow(用户界面设计工作流程)
来源仓库:https://github.com/frozenmoon/ui-design-workflow
仓库路径:skills/ui-design-workflow
安装命令:
npx skills add https://github.com/frozenmoon/ui-design-workflow --skill ui-design-workflow
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/frozenmoon/ui-design-workflow --skill ui-design-workflow

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

用于辅助界面设计、视觉规范、排版、配色、布局和交互体验优化。适合根据产品场景整理页面结构、生成 UI 方案、检查视觉一致性或改进组件层级。使用时需结合现有品牌、设计系统和用户任务,避免堆砌装饰元素;涉及页面改动时应通过截图或浏览器预览检查文本溢出、对齐和响应式表现。

安装方式:github,可通过 npx skills add 命令从指定仓库添加。适用于 Codex、Claude、Cursor、Gemini CLI 等宿主环境。

来源仓库内路径:skills/ui-design-workflow

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/frozenmoon/ui-design-workflow --skill ui-desig…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要操作浏览器,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

ui-design-workflow 怎么安装?

在终端执行 npx skills add https://github.com/frozenmoon/ui-design-workflow --skill ui-design-workflow。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

ui-design-workflow 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

ui-design-workflow 需要什么权限?

权限级别是操作浏览器。该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

ui-design-workflow 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

ui-design-workflow 用的人多吗?

公开数据是累计安装 267 次,近一周安装 11 次,下载 87 次,GitHub 3 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

ui-design-workflow 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
ui-design-workflow

用户界面设计工作流程 · 本页

2673操作浏览器未明确通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

UI Design Workflow

Manage the full UI design workflow: Demo → Spec → Implement → Check → Iterate. Supports both custom components and component libraries (shadcn/ui, Ant Design, Element Plus, etc.).

Command Reference

CommandPurposeWhen to Use
demoAnalyze project + generate UI showcaseEarly stage: establish style & component solution
specGenerate spec documentAfter Demo is confirmed
implementBuild features per specDaily development
checkAudit spec complianceBefore PR, periodic review
iterateUpdate the specWhen gaps are discovered

Command Details

demo - Generate UI Components + Showcase Page

Input: User's UI style requirements (e.g., "minimal modern", "techy")

Output:

  • Real component files in src/components/ui/ (reusable across the project)
  • Two showcase pages (separate pages, NOT tabs):

- Product Demo (/ui-showcase) - Realistic demo of the actual product, adapts to project type (see below) - Component Library (/ui-components) - All components rendered in various states, independently iterable

Flow selection (auto-detect):

  • Project has existing spec → Flow B (import existing components, generate demo)
  • User specified a library → Flow C (use specified library, still confirm style)
  • Project has significant existing UI (5+ components or 3+ pages) but no spec → Flow D (extract and document, see below)
  • Otherwise → Flow A (from scratch)

Flow D: Adopt Existing UI (for projects with existing UI code but no spec):

  1. Deep scan: Extract design tokens (colors, fonts, spacing, radius, shadows) and component inventory from existing code
  2. Present extraction results: Show findings to user via AskUserQuestion for confirmation
  3. Gap analysis: Identify missing tokens, inconsistent values, missing components
  4. User decides: Fix gaps now or defer to iterate
  5. Generate spec: doc/ui-design-spec.md based on extracted data (not invented)
  6. Optional showcase: Ask if user wants a showcase page for existing components
  7. Update agent config: Write spec reference to CLAUDE.md / AGENTS.md

Product Demo adapts to project type:

  • Website / Landing Page → Demo shows the actual website (hero, features, CTA, etc.)
  • Web App / SaaS / Dashboard → Demo shows main app screens (dashboard, list page, detail page, etc.)
  • E-commerce → Demo shows product listing, product detail, cart overview
  • Blog / Content site → Demo shows article list, article detail
  • Portfolio → Demo shows portfolio grid, project detail
  • If project type is unclear from code analysis, ask the user what they are building

Flow A (from scratch):

  1. Project Analysis + Product Understanding:

- Read package.json to identify framework and existing dependencies - Scan code directory to identify existing patterns and tech stack - Determine project type (website, web app, SaaS, e-commerce, blog, etc.) - ⛔ If project type is unclear: Ask the user what they are building (e.g., "Is this a website, a web app, a dashboard...? What are the main pages/features?"). This determines what the demo page will show. - Scan existing components: Find all components in the project, assess their consistency

  1. Present findings and auto-select component solution:

- Show existing component inventory and consistency assessment to user - Auto-select a component solution based on analysis (custom / shadcn/ui / Ant Design / other) - Auto-detect existing library dependencies and continue with them - If user already specified a library in the command, use it directly - Do NOT ask the user to choose — use the recommended solution by default. Advanced users will override if needed.

  1. ⛔ Visual Style Preview (BLOCKING) — Generate style-preview.html (standalone HTML, opens in any browser) with 3 diverse style direction previews. Each preview shows real fonts (Google Fonts), actual color palette, a mini Hero section, and atmosphere effects. User opens in browser, picks A/B/C, asks for a new batch, or describes their own direction. Then ask theme mode (light/dark/toggle). DO NOT proceed without user's confirmed choice. See assets/style-preview-template.md for the HTML template.
  2. Design Thinking (read references/frontend-aesthetics.md):

- Refine the user's chosen style direction into detailed design tokens - Define the memorable element — what makes this design unforgettable? - Finalize typography, color palette, atmospheric details — all coherent with the confirmed visual preview

  1. Create real component files (the single source of truth):

- Set up design tokens as CSS variables - Create each component as a real file in src/components/ui/ - Components support all variants, states, transitions, and accessibility - Skip components that already exist in the project

  1. Create two showcase pages (separate pages, NOT tabs):

- Product Demo (/ui-showcase): Realistic demo adapted to the project type — website shows the website, app shows main app screens, etc. Built with real components. - Component Library (/ui-components): All components in every variant/state. - Both pages share a navigation bar with links to each other. - Include existing project components in Component Library page (do not modify them) - Apply motion: Orchestrate page load animations with staggered reveals - Apply atmosphere: Add background textures, gradients, or visual depth

  1. Run anti-AI-slop checklist (see references/frontend-aesthetics.md)
  2. Iterate with user

Existing component handling:

  • Preserve existing components as-is; do not modify during demo phase
  • Only suggest fixes for severe issues (accessibility violations, broken layouts)
  • Minor inconsistencies are noted for future iterate updates

Output locations:

  • Components: src/components/ui/*.tsx
  • Product Demo (Next.js): app/ui-showcase/page.tsx | (Vite): src/pages/UIShowcase.tsx
  • Component Library (Next.js): app/ui-components/page.tsx | (Vite): src/pages/UIComponents.tsx

Templates: See assets/components.md for component library structure Library guide: See assets/library-guide.md


spec - Generate Spec Document

Prerequisite: Confirmed Demo exists (component files + showcase page)

Output:

  1. Spec document: doc/ui-design-spec.md (includes Component Registry)
  2. Project integration: Update CLAUDE.md with spec reference

Flow:

  1. Read component files and showcase code, extract design decisions
  2. Build Component Registry: Scan all component files, record name → path → variants → states
  3. Record component solution: Document the chosen solution in the spec

- Solution type (custom / library name) - Library version (if applicable) - Theme configuration method and file path

  1. Generate structured spec document (design tokens + registry, NOT code examples)
  2. Update CLAUDE.md with spec reference

implement - Build Features Per Spec

Prerequisite: Project has a UI spec with Component Registry

Flow:

  1. Must read the project's UI spec document first
  2. Check Component Registry: Find existing components that can be reused

- If a needed component exists → read its source file, import and use it - If a needed component does NOT exist → create it following the spec's design tokens, then notify user to run iterate to register it

  1. Implement the user's requested feature per spec
  2. Ensure colors, spacing, and component styles comply with the spec

Key: Always check the Component Registry before creating any UI element. Reuse existing components — do not duplicate.


check - Audit Spec Compliance

Input: File or directory to check

Flow:

  1. Read the project's UI spec (including Component Registry)
  2. Scan the specified code
  3. Check for:

- Component reuse: Are registry components being imported and used, or duplicated inline? - Design tokens: Are CSS variables used instead of hardcoded values? - Spacing/radius: Do values conform to the spec's grid system? - Library compliance (if applicable): Are library components used correctly?

  1. Output violation report with component reuse rate

iterate - Update the Spec

Trigger: New component created during implement, missing registry entry, design token change, or switching component solution

Flow:

  1. Confirm with user what needs to be added/modified
  2. If adding a new component: verify the file exists, add to Component Registry, add to showcase
  3. If switching component solution, flag components that need migration
  4. Update spec document (registry / tokens)
  5. Update Demo showcase page (if needed)
  6. Sync related implementation code (if needed)

Design Principles (General)

Full details: assets/design-principles.md — Must read before demo phase. Aesthetics guide: references/frontend-aesthetics.md — Core creative reference for all design work.

Intentional Design (Core Principle)

  • Every design must have a clear aesthetic direction and point of view
  • Choose a bold tone: brutally minimal, luxury refined, editorial, organic, retro-futuristic, etc.
  • Define the memorable element — the ONE thing someone will remember
  • NEVER produce generic AI aesthetics: no Inter/Roboto fonts, no purple-on-white gradients, no cookie-cutter layouts

Less is More (Token Principle)

  • Default to the minimum viable set of design tokens
  • Fewer named levels: border-radius 3, spacing 4, shadows 3, font weights 2
  • Minimal tokens ≠ minimal aesthetics — a small set of well-chosen tokens creates more cohesion

Component Solution

  • Custom components: Hand-write all UI components, full control
  • Component library: Use an existing library (shadcn/ui, Ant Design, Element Plus, etc.)
  • Choice is made during demo stage, recorded in the spec document
  • All subsequent stages automatically follow this choice

Typography

  • Choose distinctive, characterful fonts — never generic defaults
  • Pair a display/heading font with a refined body font for contrast
  • Vary font choices across different projects — never converge on the same "safe" pick
  • Clear type scale hierarchy (Display → H1 → H2 → Body → Small)

Color System

  • Dominant + accent model: One dominant color for mood, sharp accents for focal points
  • Build palettes that evoke an emotion — if you can't name the feeling, it's too generic
  • Avoid: purple-blue gradients, teal/coral combos, unmodified Tailwind gray scales
  • System feedback: Success, Error, Warning, Info

Motion & Atmosphere

  • Orchestrate page load with staggered reveals (highest impact)
  • Add scroll-triggered reveals, surprising hover states
  • Create atmospheric depth: gradient meshes, noise textures, radial highlights
  • Avoid: solid white/dark backgrounds with no visual interest

Responsive

  • Mobile-first design
  • Key breakpoints: sm(640px), md(768px), lg(1024px)

File Structure

ui-design-workflow/
├── SKILL.md              # This file (overview)
├── README.md             # User documentation
├── assets/
│   ├── website.md        # Landing page reference (NOT used in demo flow — AI designs freely)
│   ├── components.md     # Component library template
│   ├── style-preview-template.md  # Visual style preview page (HTML + server)
│   ├── design-principles.md  # UI design principles (references aesthetics guide)
│   ├── spec-template.md  # Spec document template
│   └── library-guide.md  # Component library integration guide
└── references/
    ├── modes.md          # Detailed mode specifications
    └── frontend-aesthetics.md  # ⭐ Core creative aesthetics guide (anti-AI-slop)

Quick Decision Guide

User says "design UI" → Use demo (analyze project, recommend component solution) User says "use shadcn/ui" → Use demo (skip recommendation, use specified library) User says "generate spec" → Use spec User says "implement feature" → Use implement (read spec first, identify component solution) User says "check code" → Use check User says "add component spec" → Use iterate User says "switch to Ant Design" → Use iterate (switch component solution)

平台分布

Codex

37.46%
按下载量换算33

Claude

29.18%
按下载量换算25

Cursor

19.24%
按下载量换算17

Gemini CLI

9.93%
按下载量换算9

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

操作浏览器

该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills