Clone — Skill para Clonar Websites em Next.js
Amanuel2x/clone-skill
Clone websites com fidelidade pixel-perfect em Next.js + Tailwind
Ver no GitHubSkill do Claude Code que clona qualquer website em um projeto Next.js 14 + Tailwind CSS pronto para produção. Extrai o sistema de design completo (CSS, animações, tipografia, assets), reconstrói como componentes React limpos e valida a fidelidade visual pixel-perfect com screenshot-diff.
Instalação
npx convez-ai Amanuel2x/clone-skill1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Recriar designs de websites concorrentes como base para rebranding
- Extrair e documentar sistemas de design de sites existentes
- Gerar scaffolds Next.js a partir de protótipos visuais ou landing pages
Veredito

Ferramenta sofisticada e bem arquitetada para engenheiros que precisam transpor designs web complexos para código React. Exige acesso legítimo ao website original e manejo cuidadoso de copyright, mas entrega fidelidade técnica genuína com pipeline robusto de verificação.
README
A tradução para o português deste README ainda está na fila.
clone — a Claude Code skill for pixel-perfect website cloning
A Claude Code skill that clones any website into a production-ready Next.js 14 + Tailwind CSS project — extracting the design system, layout, typography, assets, and every animation (scroll, hover, entrance, looping), then rebuilding it section by section as clean React components.
The goal is fidelity to the T: a clone that "looks about right" is a failed clone. The pipeline render-primes the page, captures CSS and JS-driven motion, self-hosts the exact fonts and assets, and finishes with a screenshot-diff verification loop before it calls itself done.
Note: the rebuilt site is a faithful reconstruction intended as a starting point (e.g. rebranding for a client). Only clone sites you have the right to copy. Respect copyright, trademarks, and licensing.
What it does
/clone https://example.com for acme-corp
Running the skill executes a 12-step pipeline:
- Parse the command → source URL + client slug + target directory.
- Map the site with Firecrawl to discover all pages.
- Extract the design system — every CSS variable,
@keyframes(verbatim),@font-facerule, per-role font metrics, color palette by usage, gradients, shadows, and the spacing/radius/z-index scales. - Capture dynamic animations — detect GSAP / Framer Motion / AOS / Lenis / Swiper / Locomotive and build a motion manifest of what triggers each animation, its transform deltas, duration, easing, and stagger.
- Scrape the full rendered HTML (Firecrawl + a live-browser DOM cross-check).
- Screenshot every section at desktop, tablet, and mobile widths.
- Download all assets and self-host the exact fonts — no hotlinking, no "close enough" substitutions.
- Scaffold a Next.js 14 App Router project (TypeScript + Tailwind + Framer Motion).
- Generate
tailwind.config.ts,globals.css,lib/animations.ts, andconfig/client.tsfrom the extracted tokens. - Build each section as a React component, matching the original's box model and animations.
- Register the client and run it locally.
- Verify fidelity — screenshot-diff the rebuild against the original at every breakpoint and iterate until parity.
Output structure
Each clone produces a self-contained Next.js project:
<client-slug>/
├── app/
│ ├── layout.tsx # Root layout — fonts (next/font) + metadata
│ ├── page.tsx # Homepage assembles all sections in order
│ └── globals.css # CSS variables + @font-face + @keyframes (verbatim)
├── components/
│ ├── sections/ # One file per section (Hero, Features, …)
│ ├── ui/ # Reusable UI: Button, Card, Badge, …
│ └── layout/ # Header, Footer, Nav
├── config/
│ └── client.ts # ALL client-specific content lives here
├── lib/
│ └── animations.ts # Framer Motion variants from keyframes + motion manifest
├── public/
│ └── assets/ # Downloaded logo, images, fonts, videos
└── tailwind.config.ts # Extended with extracted colors/fonts/spacing/radius/shadows
All swappable content is isolated in config/client.ts, so the same clone can be re-skinned for a new brand without touching components.
Requirements
| Requirement | Why |
|---|---|
| Claude Code | The runtime that executes the skill. |
| Playwright MCP | Renders the live site, runs in-page JS, captures animations and screenshots. Required. |
| Firecrawl MCP | Maps the sitemap and scrapes JS-rendered HTML. Recommended (degrades to homepage-only without it). |
| Node.js 18+ | For create-next-app and running the generated project. |
This pipeline uses Firecrawl for scraping — not Firebase. Firebase is a database/hosting product and plays no role in cloning.
Install the MCP servers
Playwright:
claude mcp add playwright -- npx -y @playwright/mcp@latest
npx -y playwright install chromium # if screenshots report "browser not installed"
Firecrawl (get a free key at https://www.firecrawl.dev):
claude mcp add firecrawl --env FIRECRAWL_API_KEY=fc-YOUR_KEY_HERE -- npx -y firecrawl-mcp
Verify both are connected:
claude mcp list # → playwright and firecrawl should show ✓ Connected
Installing the skill
Drop SKILL.md into a clone/ folder inside your Claude Code skills directory:
mkdir -p ~/.claude/skills/clone
curl -sL https://raw.githubusercontent.com/Amanuel2x/clone-skill/master/SKILL.md \
-o ~/.claude/skills/clone/SKILL.md
Or clone this repo and copy the file:
git clone https://github.com/Amanuel2x/clone-skill.git
cp clone-skill/SKILL.md ~/.claude/skills/clone/SKILL.md
Restart Claude Code (or start a new session) and the /clone skill will be available.
Usage
/clone https://example.com # clones, slug derived from the domain
/clone https://example.com for acme-corp # clones into the "acme-corp" client slug
clone this site: https://example.com # natural-language trigger
After a clone completes:
- Drop the client's real assets into
public/assets/(logo.svg,hero.jpg,brand-colors.json,copy.json). - Re-skin with a companion
swap-assetsstep (if you use one). cd <client-slug> && npm run dev.
How fidelity is enforced
Most "website cloners" screenshot a half-loaded page and approximate the styling. This skill is built to avoid that:
- Render-priming — waits for
document.fonts.readyand scrolls the full page height so lazy-loaded images and scroll-triggered animations fire before anything is captured. - Verbatim keyframes —
@keyframesand@font-facerules are copied in full, never truncated. - Motion manifest — JS-driven animations (which static CSS extraction misses) are detected and documented with their triggers, deltas, durations, and easings, then recreated as Framer Motion variants.
- Exact fonts — the original typefaces are self-hosted (via
next/fontor copiedwoff2files); typeface substitution is the single biggest fidelity killer and is explicitly avoided. - Multi-viewport capture — every section is screenshotted at desktop, tablet, and mobile.
- Verification loop — the rebuild is screenshot-diffed against the original top-to-bottom and iterated until it matches.
License
MIT — see LICENSE.