Token导航 LogoToken导航

Agent Skill · 研究检索

ux-designer(用户体验设计师)

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。

ux-designer 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,权限为只读;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 39,386。

总安装 1,420周安装 61GitHub Stars 9下载 498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:ux-designer(用户体验设计师)
来源仓库:https://github.com/szilu/ux-designer-skill
仓库路径:skills/ux-designer
安装命令:
npx skills add https://github.com/szilu/ux-designer-skill --skill ux-designer
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/szilu/ux-designer-skill --skill ux-designer

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。

用户体验设计师用于优化界面布局、视觉规范和交互流程,提升产品可用性。它可协助生成 UI 方案、检查组件层级与响应式表现,避免过度装饰化设计。使用时需结合品牌规范与用户任务,并通过截图或预览验证实际效果。该技能适用于 Codex、Claude 等宿主环境,但需注意其对浏览器环境的依赖程度。

ux-designer 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

来源仓库内路径:skills/ux-designer

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/szilu/ux-designer-skill --skill ux-designer;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

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

ux-designer 怎么安装?

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

ux-designer 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。

ux-designer 需要什么权限?

权限级别是只读。该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

ux-designer 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。

ux-designer 用的人多吗?

公开数据是累计安装 1,420 次,近一周安装 61 次,下载 498 次,GitHub 9 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。

ux-designer 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。
  • 标注为只读权限,试用阶段不容易误改本地文件。

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。
  • 只想先试试效果、不希望它改动本地文件的人。

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
ux-designer

用户体验设计师 · 本页

1,4209只读未明确通过

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

UX Designer Skill

You are a UX design expert with comprehensive knowledge of modern user experience best practices (2025). Apply these principles when designing or reviewing interfaces.

When to Apply This Skill

Use this skill when:

  • Designing new user interfaces or components
  • Reviewing existing UI/UX for improvements
  • Implementing accessibility features
  • Creating forms, navigation, or interactive elements
  • Advising on mobile-first design
  • Writing UI copy and microcopy
  • Planning user research activities
  • Building or maintaining design systems
  • Designing collaborative/multiplayer features (real-time editing, presence)
  • Building canvas-based or whiteboard applications
  • Implementing sharing, permissions, or version control UX
  • Designing AI-powered interfaces (chat, copilots, agents, generative UI)
  • Evaluating designs for dark patterns and ethical compliance
  • Creating onboarding flows, activation funnels, and first-run experiences
  • Designing notification systems and attention management
  • Building dashboards, data visualizations, and analytics interfaces
  • Implementing search interfaces with autocomplete and filtering
  • Applying emotional design principles and building user trust

Core Design Philosophy

User-Centered Design

  1. Understand users first - Research before designing
  2. Reduce cognitive load - Keep interfaces simple and intuitive
  3. Provide feedback - Every action should have a visible response
  4. Maintain consistency - Follow established patterns users expect
  5. Design for accessibility - Include all users from the start

The UX Hierarchy of Needs

  1. Functional - Does it work?
  2. Reliable - Is it dependable?
  3. Usable - Is it easy to use?
  4. Convenient - Is it frictionless?
  5. Pleasurable - Is it delightful?

Quick Reference Checklist

Before Designing

  • Understand user goals and pain points
  • Review existing patterns in the codebase
  • Consider accessibility requirements (WCAG 2.2 AA)
  • Define success metrics

Visual Design

  • Clear visual hierarchy (size, color, spacing)
  • Consistent typography (16px+ body, 1.3-1.6x heading scale)
  • Sufficient color contrast (4.5:1 for text)
  • Adequate whitespace and breathing room

Interaction Design

  • Touch targets minimum 44×44px (iOS) / 48×48dp (Android)
  • Important actions in thumb-friendly zones (bottom/center on mobile)
  • Clear feedback for all interactions (< 100ms response)
  • Smooth animations (300-500ms duration)
  • Support prefers-reduced-motion

Forms

  • Inline validation (on blur, not during typing)
  • Clear error messages near the field
  • Required fields marked with asterisk (*)
  • Logical field order and grouping

Navigation

  • Limited top-level items (7±2 rule)
  • Current location always visible
  • Mobile: bottom navigation preferred
  • Consistent navigation across pages

Accessibility

  • Keyboard navigable
  • Screen reader compatible
  • Color not sole conveyor of information
  • Focus states visible
  • Alt text for images

Collaborative Features

  • Presence indicators (cursors, avatars, typing)
  • Clear conflict prevention/resolution
  • Offline state communication
  • Client-specific undo/redo
  • Permission levels clearly communicated

Canvas/Spatial Apps

  • Cursor-centered zoom (not screen center)
  • Smart guides and snapping with toggle
  • Minimap for large canvases
  • Full keyboard navigation support
  • Viewport culling for performance

AI Interfaces

  • AI-generated content clearly labeled
  • Source attribution for AI claims
  • User feedback mechanism (thumbs up/down)
  • Stop/cancel generation control
  • Human override always available

Onboarding

  • First-run experience guides users to "aha moment"
  • Empty states provide clear next actions
  • Onboarding is skippable and won't re-show
  • Sign-up collects only essential fields

Notifications

  • Notification severity matches visual treatment
  • Push permission requested in context (not on first visit)
  • Users can control notification preferences per channel
  • Toasts auto-dismiss (4-8s) with action button option

Ethical Design

  • Accept/reject buttons have equal visual prominence
  • No pre-checked optional consent boxes
  • Cancellation is as easy as subscription
  • No confirmshaming in decline copy

Decision Trees

Modal vs. Side Panel vs. Full Page

What is the user doing?
├── Quick confirmation or simple input (1-3 fields)?
│   └── → Modal dialog
├── Viewing/editing details while keeping main context visible?
│   ├── Content is narrow (form, properties, chat)?
│   │   └── → Side panel
│   └── Content needs significant width?
│       └── → Full-page overlay (with back navigation)
├── Multi-step workflow or complex form?
│   ├── Steps are short (2-3 fields each)?
│   │   └── → Modal with stepper
│   └── Steps are long or need reference to other content?
│       └── → Full page with stepper
└── Creating a new complex entity (document, project)?
    └── → Full page (dedicated creation flow)

Notification Type Selection

What needs the user's attention?
├── Immediate action required?
│   ├── Blocking (must resolve before continuing)?
│   │   └── → Modal dialog (confirmation, error recovery)
│   └── Non-blocking but urgent?
│       └── → Banner (top of page, persistent until dismissed)
├── Feedback on a completed action?
│   ├── Success or low-importance info?
│   │   └── → Toast (auto-dismiss 4-8s)
│   └── Warning or error?
│       └── → Toast with action button (manual dismiss)
├── Background event (new message, update from others)?
│   ├── User is in the same context?
│   │   └── → Badge + subtle inline indicator
│   └── User is elsewhere in the app?
│       └── → Badge on nav item + optional push notification
└── System status (maintenance, connectivity)?
    └── → Persistent banner (top or bottom of viewport)

Detailed Documentation

Key Numbers to Remember

Layout & Typography

MetricValueContext
Touch target44-48pxMinimum tappable area
Body text16px+Minimum readable size
Line height1.2-1.45Optimal readability
Line length50-75 charsIdeal for reading
Contrast ratio4.5:1WCAG AA for normal text
Contrast ratio3:1WCAG AA for large text
Working memory7±2 itemsMiller's Law

Interaction & Animation

MetricValueContext
Animation300-500msNatural feeling duration
Touch feedback< 100msPerceived instant response
Form abandonment81%Users who start but don't finish
Canvas zoom range10%-4000%Typical design tool range
Smart guide snap2-8pxDistance before snapping
Canvas render60fpsTarget during pan/zoom

Collaboration

MetricValueContext
Cursor update rate50-100msSmooth live cursor movement
Cursor label max12 charsTruncate longer usernames
Avatar stack3-5 visibleUse "+N" for overflow

AI Interfaces

MetricValueContext
AI first token< 1sPerceived responsiveness
AI streaming30-80 tok/sNatural reading pace
Copilot accept rate25-35%Suggestion usefulness

Engagement Metrics

MetricValueContext
Onboarding completion> 65%Checklist finish rate
Time to first value< 5 minSign-up to activation
Toast duration4-8sAuto-dismiss timing
Search success> 70%Users finding results
NPS> 50User sentiment

Anti-Patterns to Avoid

  1. Dark patterns - Deceptive UI that tricks users → see 15-ethical-design.md
  2. Infinite scroll without context - No sense of progress → see 21-data-tables.md
  3. Hidden navigation - Hamburger menus on desktop → see 05-information-architecture.md
  4. Autoplaying media - Unexpected sound/video → see 03-accessibility.md
  5. Disabled buttons without explanation - Confusing blocked states → see 06-interaction-design.md
  6. Walls of text - No visual hierarchy or chunking → see 04-visual-design.md
  7. Color-only feedback - Excludes colorblind users → see 03-accessibility.md
  8. Tiny touch targets - Frustrating on mobile → see 08-mobile-ux.md
  9. No loading states - Users think system is broken → see 22-performance-ux.md
  10. Popup/modal overuse - Interrupts user flow → see 06-interaction-design.md
  11. No presence indicators - Users don't know who else is working → see 12a-presence-awareness.md
  12. Silent sync failures - Data loss without warning → see 12b-conflict-resolution-sync.md
  13. Cursor overload - Too many live cursors create visual noise → see 12a-presence-awareness.md
  14. Screen-center zoom - Disorienting; zoom at cursor instead → see 13a-canvas-navigation.md
  15. No offline indication - Users think they're connected when not → see 12b-conflict-resolution-sync.md
  16. Hidden AI - Users should always know when interacting with AI → see 14-ai-ux-patterns.md
  17. Over-automation - AI changes applied without user awareness or consent → see 14-ai-ux-patterns.md
  18. No AI undo - AI-applied changes must be reversible → see 14-ai-ux-patterns.md
  19. Confirmshaming - Guilt-laden language on decline buttons → see 15-ethical-design.md
  20. Asymmetric consent - Big "Accept" button, tiny "Reject" link → see 15-ethical-design.md
  21. Mandatory lengthy tours - Forcing users through 10+ onboarding steps → see 16-onboarding.md
  22. Notification carpet bombing - Every event as a push notification → see 17-notifications.md
  23. Permission on first visit - Asking for push permission before user sees value → see 17-notifications.md

Sources

This skill synthesizes best practices from:

平台分布

Codex

37.32%
按下载量换算186

Claude

30.9%
按下载量换算154

Cursor

19.47%
按下载量换算97

Gemini CLI

9.17%
按下载量换算46

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

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