Help us improve
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
By kkunkunya
Guides frontend design and development through a structured AI-driven workflow: from requirement analysis and design research to visual asset generation, code implementation prompts, and quality reviews. Includes visual rules audits, design system documentation, i18n guidance, incremental change routing, and motion animation prompts.
npx claudepluginhub kkunkunya/ai-frontend-design-kit --plugin ai-frontend-design-kit前端代码生成前或评审时的反 slop 硬门禁。扫十条作者硬判断的视觉红线——Inter 字体、紫/蓝 AI 渐变(Lila Ban)、3 列等宽卡片、纯黑
阶段 2 试吃——对 4-5 个该领域设计优秀的真实网站跑六维机械拆解(形 / 色 / 字 / 构 / 质 / 动),综合成本项目 design.md 初稿。六维是上位骨架,DOM 量化工作流 + 动效采访 5 类是落地肌肉。每次一站,保护上下文。不做前轮/后轮深度采访(那是 frontend-interview-dualround)、不做视觉锚定三件套(那是 frontend-visual-reference)、不写最终交付的 design.md(那是 frontend-design-writer)、不做组件库 / 动效资产库选型(那部分由 frontend-design-writer 的 §11 Resources 段 + 项目级 components.json 决定)。触发:新项目六维调研、整站重做、前端方向大迭代。
代码已经跑起来后(阶段 7 迭代微调)的设计评审横切——用户看到具体页面涌现审美反馈时使用:三段式追问(定位→对标→量化)把"感觉不对"翻译成可执行改动项,"选择题化"让用户做选择题而不是填空题,参数外显 Tweaks 让不确定的参数暴露成可调旋钮,所有修正反写回 design.md。覆盖 design-review(视觉对标)、design-content(文案与信息层级)、design-harden(边缘状态加固)三种子模式。**仅用于代码已生成后**——代码前的"感觉"引导走 frontend-interview-dualround 的后轮采访;纯调研走 frontend-design-research;写 design.md 走 frontend-design-writer。
阶段 1-7 持续维护的 design.md 主写手——把调研(六维试吃)+ 双轮深度采访 + 视觉锚定三件套 + 动效实现 prompt 集全部压缩为 Agent 可执行的设计契约。kun 版 15 段三轮写作顺序(原 12 段 + 新增 §12 Moodboard Prompt / §13 Page Mockup Prompt / §14 Motion Prompts 三段视觉锚定引用)。同时维护 DESIGN-LANGUAGE.md(低频)+ design.md(高频)+ MOTION-SPEC.md(中频)三份。不用于调研本身(那是 frontend-design-research)、不用于从零搭 knowledge/ 骨架(那是 req-suite:project-spec)、不用于单 token 微调(直接编辑即可)、不用于生成微观动效实现 prompt(那是 frontend-motion-prompt-writer)。
前端国际化(i18n)的综合规划与执行指南——做国际性网站、多语言架构、补多语言、翻译质量验证、字体跨语言统一、路由 / SEO / 字符串提取 / 动态切换等 i18n 相关任务时用。本 skill 按"运行时 / 翻译内容 / 视觉"三层正交模型路由:先定位用户的 i18n 痛点在哪一层,再走对应决策流;翻译内容层默认走三段流水线(机翻 → AI 本地人审 → 人工兜底)+ subagent 并行化(locale × persona × domain)。触发:用户说"做国际性网站""加多语言""补英文版""做 i18n""多语言网站的 SEO""路由用子路径还是子域名""本地化翻译""文化陷阱""AI 扮演当地人审翻译""字体跨语言不统一""语言切换器怎么做""字符串提取"。不用于纯单语言产品(没有国际化意图的就别套)、不用于纯翻译任务不涉及前端(那是纯文案)、不用于 RTL 排版深度实现(附录里有清单但不是本 skill 主场,目标市场不含中东时直接跳过)。
Share bugs, ideas, or general feedback.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Memory compression system for Claude Code - persist context across sessions
Ultra-compressed communication mode. Cuts ~75% of tokens while keeping full technical accuracy by speaking like a caveman.
Frontend design skill for UI/UX implementation
Comprehensive UI/UX design plugin for mobile (iOS, Android, React Native) and web applications with design systems, accessibility, and modern patterns
Standalone image generation plugin using Nano Banana MCP server. Generates and edits images, icons, diagrams, patterns, and visual assets via Gemini image models. No Gemini CLI dependency required.
Marketing skills for AI agents — conversion optimization, copywriting, SEO, paid ads, ad creative, and growth
PPT production toolkit: route presentation requests across HTML showcase, image assets, editable PPTX, fixed-template fill, and verification.
Experiment planning toolkit for P4 academic research: turn claims into evidence chains, baseline/ablation plans, figure/table specs, feasibility salvage paths, and req-suite handoff tasks.
This is a portfolio-facing Agent/Plugin project for frontend design direction, design constraints, and Claude Code/Codex handoff workflows. It demonstrates how I package reusable AI engineering capability as installable project tooling.
English · 中文
Constraints before generation. A 10-skill plugin for Claude Code and Codex that upgrades AI-assisted frontend design from "generate four options, pick one" to a rigorous 7-stage flow with taste-level design contracts.
Most "AI frontend" tooling generates 2–4 page variants and asks you to pick one. That works for throwaway demos. It does not work when you actually care about the result — because AI cannot form your aesthetic judgment for you.
This kit inverts the loop:
The result is a persistent design.md (kun 15-segment format) that every subsequent AI agent reads before touching a single line of JSX.
# Add this repo as a marketplace
claude plugin marketplace add https://github.com/kkunkunya/ai-frontend-design-kit
# Install the plugin
claude plugin install ai-frontend-design-kit@ai-frontend-design-kit
# Clone anywhere
git clone https://github.com/kkunkunya/ai-frontend-design-kit.git ~/codex-plugins/ai-frontend-design-kit
# Register in ~/.codex/config.toml
[[plugins]]
source = "~/codex-plugins/ai-frontend-design-kit"
enabled = true
After install, ask Claude Code or Codex:
Start a new frontend project, use the 7-stage flow.
The agent should route to frontend-interview-dualround (stage 1). If it does, you're set.
| Skill | Stage | One-line job |
|---|---|---|
frontend-iteration-planner | Scenario C entry | Increment router — read existing design.md + assets, then tier the change into T1 new-page / T2 new-section / T3 micro-tweak |
frontend-interview-dualround | Cross-cut (stage 1 + 3) | Product-designer perspective five-class interview, pre-round to set boundaries + post-round to extract details |
frontend-design-research | Stage 2 | Six-dimension tasting (form / color / type / layout / material / motion) across 4–5 reference sites → design.md draft |
frontend-visual-reference | Stage 4 | Visual anchor triptych (moodboard exploration + page mockup + motion frame sequence) via baoyu-imagine + manual fallback |
frontend-motion-prompt-writer | Stage 4C + 6 | Translates MOTION-SPEC + frame sequence into LoadingScreen-style code-level prompts |
frontend-design-writer | Stages 1–7 | The main writer. Kun 15-segment 4-round workflow; maintains design.md + MOTION-SPEC.md + DESIGN-LANGUAGE.md |
frontend-design-review | Stage 7 | Post-code three-stage interrogation + Tweaks parameter externalization + write-back to design.md |
frontend-anti-slop-gate | Stage 6 cross-cut | Ten hard red-lines scan (Inter font / Lila palette / 3-column grid / h-screen / placeholder data / etc.) |
system-design | Orthogonal | System architecture: Design Doc + ADR + structural-impact analysis |
frontend-i18n-essentials | Orthogonal | i18n three-layer orthogonal model (runtime / translation-content / visual) + subagent-parallel translation pipeline |
Stage 1 Requirement + pre-round interview ── frontend-interview-dualround (pre)
Stage 2 Six-dimension tasting × 4-5 sites ── frontend-design-research
Stage 3 Post-round interview ── frontend-interview-dualround (post)
Stage 4 Visual anchor triptych ── frontend-visual-reference (A moodboard + B mockup + C motion frames)
Stage 5 Asset generation (dual path) ── baoyu-imagine primary + manual fallback
Stage 6 Code implementation ── your usual dev flow + anti-slop gate
Stage 7 Iteration and micro-tuning ── frontend-design-review
Cross-cuts:
├─ Anti-slop hard bans (stage 6 resident) ── frontend-anti-slop-gate
└─ Dualround interview (stages 1 + 3) ── frontend-interview-dualround
Orthogonal:
├─ System architecture decisions ── system-design
└─ i18n planning (multilingual sites) ── frontend-i18n-essentials
Throughline: design.md is continuously maintained ── frontend-design-writer (kun 15 segments)