Token导航 LogoToken导航

Agent Skill · 前端设计

design-first-dev(设计第一开发)

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。

design-first-dev 是前端设计类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为需要联网;许可证信息还不明确,第三方扫描已通过,按安装量在前端设计类的 10,216 个 Skill 里排第 7,463。

总安装 372周安装 16GitHub Stars 317下载 131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:design-first-dev(设计第一开发)
来源仓库:https://github.com/menkesu/awesome-pm-skills
仓库路径:skills/design-first-dev
安装命令:
npx skills add https://github.com/menkesu/awesome-pm-skills --skill design-first-dev
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/menkesu/awesome-pm-skills --skill design-first-dev

这个 Skill 是做什么的

前端设计类 Skill 关注页面结构、视觉规范和组件写法,输出通常是可运行的界面代码。

推行 Airbnb 式设计主导开发模式,强调细节把控与用户体验优先。适合制定界面规范、核心流程设计及关键节点体验打磨策略。区分微管理与细节关注,确保团队在正确方向上高效执行。需建立量化评估标准,避免主观偏好影响客观质量判断。design-first-dev 属于前端设计类 Skill,可作为该场景下的辅助能力补充。

来源仓库内路径:skills/design-first-dev

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/menkesu/awesome-pm-skills --skill design-first…;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

    它需要需要联网,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

design-first-dev 怎么安装?

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

design-first-dev 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。

design-first-dev 需要什么权限?

权限级别是需要联网。该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

design-first-dev 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。

design-first-dev 用的人多吗?

公开数据是累计安装 372 次,近一周安装 16 次,下载 131 次,GitHub 317 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。

design-first-dev 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。
  • 需要处理前端设计类任务、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。

前端设计类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
frontend-design

前端设计 · Codex、Claude

8,823,360126,072只读MIT通过
vercel-composition-patterns

Vercel 组合模式 · Codex、Claude

3,858,81625,893需要联网MIT通过
vercel-react-native-skills

Vercel React native skills 前端 · Codex、Claude

2,468,66425,878只读MIT通过
adapt

adapt 搜索 · Codex、Claude

2,027,04023,125操作浏览器未明确通过
bolder

bolder 分析 · Codex、Claude

2,004,28823,130只读未明确通过
design-first-dev

设计第一开发 · 本页

372317需要联网未明确通过

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

Design-Led Development: Craft Quality Standards

When This Skill Activates

Claude uses this skill when:

  • Building user-facing features or interfaces
  • Making UI/UX decisions
  • Deciding between quick prototype vs polished experience
  • Creating onboarding, core flows, or key moments
  • Determining if "details matter" for this feature

Core Frameworks

1. Airbnb's Design-Led Process (Source: Brian Chesky, CEO of Airbnb)

Leaders in the Details:

"There's a difference between micromanagement and being in the details. If you don't know the details, how do you know people are doing a good job? I made sure I was in the details."

The Approach:

🎨 Design Complete Flow First

  • Prototype full experience before writing code
  • Show, don't tell (working prototype > requirements doc)
  • Get design right, then build it right
  • Every feature needs a story (how you'll talk about it)

All States Matter:

  • Loading states (what user sees while waiting)
  • Error states (graceful failures)
  • Empty states (first-time user experience)
  • Success states (celebrations, confirmations)

Cross-Functional from Day 1:

  • Design, engineering, marketing together
  • Not sequential (design → eng → marketing)
  • Concurrent: everyone contributes to the prototype

How to Apply:

DON'T:
- Write code first, design later
- "We'll polish it after we ship"
- Design only happy path

DO:
- Design complete experience (all states)
- Prototype before building
- Include cross-functional input early
- Craft the details that users notice

Example:

Feature: "User onboarding flow"

Design-First Approach:
✅ Prototype full flow in Figma/code sandbox
✅ Include:
   - Welcome screen (first impression)
   - Loading states (fetching user data)
   - Empty state (no content yet)
   - Error state (setup failed)
   - Success state (celebration!)
   - First value moment (aha!)
✅ Show to team before building backend
✅ Iterate on prototype, then build

2. Figma's Craft Quality Philosophy (Source: Dylan Field, CEO of Figma)

AI Makes Craft the Moat:

"AI makes design, craft, and quality the new moat for startups. The bar for quality is going way up. Craft quality is how you differentiate."

When Details Matter:

🎯 Details Create Moats When:

  1. Core Product Experience - The main value loop
  2. First Impressions - Onboarding, landing pages, signup
  3. Frequent Use - Features used daily
  4. Brand Differentiation - What makes you unique
  5. Competitive Advantage - Where quality = conversion

Move Fast When:

  1. Internal Tools - Team-facing, not customer-facing
  2. Experiments - Testing hypotheses quickly
  3. Non-Core Features - Support features, admin panels
  4. Behind the Scenes - Logging, monitoring, ops

How to Apply:

Ask: "Does craft quality matter here?"

HIGH CRAFT (polish details):
- User-facing core flows
- Onboarding experiences
- Key conversion moments
- Brand touchpoints
- Competitive differentiators

LOW CRAFT (move fast):
- Internal dashboards
- Admin panels
- Quick experiments
- Support tooling
- Behind-the-scenes

Craft Quality Checklist:

  • Consistent spacing (8px grid)
  • Proper hierarchy (typography scale)
  • Smooth interactions (animations, transitions)
  • Responsive (works on all screen sizes)
  • Accessible (keyboard nav, screen readers)
  • Loading states (skeleton screens, spinners)
  • Error handling (helpful messages)
  • Empty states (guide to first value)

3. The One Roadmap Principle (Source: Brian Chesky)

Coherent Product, Not Feature Salad:

"We shifted to one company roadmap. This meant we could have a coherent story. Every feature connects to a narrative."

Product Coherence:

🎭 The Story Test:

  • Can you tell a story about how all features connect?
  • Would a customer understand the vision?
  • Is this additive to the narrative or distracting?

How to Apply:

Before building any feature:
1. How does this fit the product story?
2. Does this reinforce the core value prop?
3. Will users understand why this exists?
4. Can we talk about this in one coherent launch?

BAD: Random feature that "users requested"
GOOD: Feature that advances the product narrative

Example:

Product: "Project management tool"
Story: "See everything, finish anything"

Feature Ideas:
✅ Timeline view (fits story: "see everything")
✅ Task dependencies (fits story: "finish anything")
❌ Chat feature (distracts from story)
❌ Time tracking (doesn't reinforce core value)

Coherence test: Do new features strengthen the story?

4. Design System Thinking

Build Once, Use Everywhere:

When to Invest in Design System:

  • Building 3+ similar components
  • Multiple teams/products need consistency
  • Brand consistency is important
  • Onboarding new designers/engineers

Design System Basics:

FOUNDATIONS:
- Colors (primary, secondary, grays, feedback)
- Typography (scale, weights, line heights)
- Spacing (8px grid: 4, 8, 16, 24, 32, 48, 64)
- Radius (corners: 4, 8, 16)
- Shadows (elevation levels)

COMPONENTS:
- Buttons (primary, secondary, ghost)
- Inputs (text, select, checkbox, radio)
- Cards, Modals, Tooltips
- Navigation patterns
- Feedback (alerts, toasts, loading)

PATTERNS:
- Forms (layout, validation, submission)
- Tables (sorting, filtering, pagination)
- Empty states, Error states, Loading states

Decision Tree: When to Polish vs Ship

FEATURE: [Ready to build]
│
├─ Is this user-facing? ───────────────┐
│  YES (customer sees it) ─────────────┤
│  NO (internal) ──────────────→ MOVE FAST
│
├─ Is this core product experience? ───┤
│  YES (main value loop) ──────→ HIGH CRAFT
│  NO (supporting feature) ────────────┤
│
├─ Is this first impression? ──────────┤
│  YES (onboarding, signup) ───→ HIGH CRAFT
│  NO (later in journey) ───────────────┤
│
├─ Used frequently? ────────────────────┤
│  YES (daily/weekly) ──────────→ HIGH CRAFT
│  NO (occasionally) ───────────────────┤
│
├─ Competitive differentiator? ────────┤
│  YES (unique to us) ───────────→ HIGH CRAFT
│  NO (table stakes) ───────────→ GOOD ENOUGH
│
└─ DECISION ←──────────────────────────┘

Action Templates

Template 1: Design-First Feature Spec

# Feature: [Name]

## The Story
How this fits the product narrative:
- Connection to core value: [explain]
- User story: "[how we'll talk about this]"

## Design-First Approach

### 1. Prototype First
- [ ] Full flow designed (not just happy path)
- [ ] All states included (loading, error, empty, success)
- [ ] Prototype reviewed with team
- [ ] Clickable demo ready

### 2. States to Design
- [ ] **Loading:** [what users see while waiting]
- [ ] **Error:** [graceful failure + recovery]
- [ ] **Empty:** [first-time experience + guidance]
- [ ] **Success:** [confirmation + next steps]

### 3. Craft Quality Bar
**This feature is:** [core product / supporting / internal]
**Craft level:** [high / medium / low]

**If HIGH craft:**
- [ ] Consistent spacing (8px grid)
- [ ] Typography hierarchy clear
- [ ] Smooth interactions
- [ ] Responsive design
- [ ] Accessible
- [ ] Delight moments

### 4. Cross-Functional Input
- [ ] Design reviewed
- [ ] Engineering reviewed (feasibility)
- [ ] Marketing input (how to talk about it)
- [ ] Product narrative alignment

Template 2: State Design Checklist

For every user-facing feature:

StateQuestionDesigned?
LoadingWhat does user see while fetching data?[]
ErrorWhat if API fails? Network error?[]
EmptyWhat if user has no data yet?[]
SuccessHow do we confirm action completed?[]
First UseWhat does new user see?[]
PartialWhat if data is incomplete?[]

Template 3: Craft Quality Assessment

# Feature: [Name]

## Craft Quality Decision

### Context
- User-facing: [yes/no]
- Core product loop: [yes/no]
- First impression: [yes/no]
- Usage frequency: [daily/weekly/monthly/rarely]
- Competitive differentiator: [yes/no]

### Craft Level: [HIGH / MEDIUM / LOW]

**If HIGH Craft:**
- Investment: [X days for polish]
- Focus areas: [list what makes it special]
- Success: [what "great" looks like]

**If LOW Craft:**
- Ship threshold: [works, looks okay]
- Time budget: [X days max]
- Polish later: [yes/no]

Quick Reference Card

🎨 Design-First Checklist

Before You Code:

  • Prototype complete experience (not just happy path)
  • All states designed (loading, error, empty, success)
  • Craft level determined (high/medium/low)
  • Fits product narrative (story test passed)
  • Cross-functional input gathered

During Build:

  • Building what was designed (no "I'll fix it later")
  • Maintaining craft standards if high-craft feature
  • Using design system components
  • Testing all states work

Before Ship:

  • All states implemented
  • Craft quality matches requirements
  • Responsive on all devices
  • Accessible
  • Story ready (how to talk about it)

Real-World Examples

Example 1: Airbnb's Product Redesign (Brian Chesky)

Challenge: Rebuild entire product to be more coherent

Design-First Approach:

  • Created one company roadmap (not 50 team roadmaps)
  • Designed full experience before building any piece
  • Brian stayed in every design detail
  • Every feature had to fit the narrative

Result:

  • Coherent product launch
  • Clear story customers understood
  • Features that reinforced each other

Example 2: Figma's Core Canvas (Dylan Field)

Decision: Craft quality on core editing experience

High Craft Investment:

  • 60fps canvas rendering (smooth = moat)
  • Pixel-perfect precision
  • Multiplayer cursors (delight moment)
  • Keyboard shortcuts (pro user love)

Result:

  • Best-in-class editing experience
  • Users switch from competitors for "feel"
  • Craft quality = competitive advantage

Example 3: Stripe's Developer Experience

Decision: Polish docs and onboarding

Design-First:

  • Designed empty state (first API call)
  • Created guided tutorials
  • Perfected error messages
  • Made docs beautiful

Result:

  • Fastest developer onboarding in payments
  • Design quality = trust signal
  • Higher conversion from trial to paid

Common Pitfalls

❌ Mistake 1: Polish Everything

Problem: Treating internal tools like customer-facing products Fix: Reserve high craft for features where quality = moat

❌ Mistake 2: Ship Happy Path Only

Problem: Forgot loading/error/empty states Fix: Design all states before building any

❌ Mistake 3: Design After Building

Problem: "We'll polish it later" (never happens) Fix: Prototype first, build second

❌ Mistake 4: Feature Salad

Problem: Adding features that don't fit narrative Fix: Story test - does this strengthen core message?


Related Skills

  • zero-to-launch - For MVP scoping with design lens
  • quality-speed - For deciding when to polish vs ship
  • strategic-build - For knowing if this is high-leverage work
  • ai-product-patterns - For AI-specific UX patterns

Key Quotes

Brian Chesky:

"Leaders are in the details. The question isn't whether to be in the details, but which details matter."

Dylan Field:

"With AI, everyone can build. The differentiator is craft. Quality is the new moat."

Brian on Coherence:

"You can have 50 marketing efforts but no customer heard anything. That's because there's no coherent story."

Further Learning

  • references/airbnb-design-process.md - Full design-led methodology
  • references/figma-craft-standards.md - Craft quality patterns
  • references/design-system-starter.md - Quick start guide
  • references/all-states-examples.md - Loading, error, empty, success patterns

平台分布

Codex

34.99%
按下载量换算46

Claude

32.51%
按下载量换算43

Cursor

19.9%
按下载量换算26

Gemini CLI

10.07%
按下载量换算13

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

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