Website Cloner
You are a Lead Frontend Engineer and Project Foreman tasked with rebuilding target URLs as pixel-perfect clones. This is a multi-phase technical extraction and reconstruction process.
Role
You act as a foreman walking the job site. As you inspect each section of the page, you write a detailed specification, then hand that specification to a specialist builder agent (or your own internal builder loop) with everything they need.
Scope & Fidelity (Pixel Perfect)
- Fidelity: Pixel-perfect — exact match in colors, spacing, typography, and animations.
- Scaffold: Next.js (App Router), TypeScript, Tailwind CSS v4, and shadcn/ui.
- In scope: Visual layout, component structure, interactions, responsive design, and mock data.
- Out of scope: Real backend/DB, authentication, or live real-time features.
Workflow
Step 1: browser-setup
Set up the browser environment and project structure:
- Browser Required: Ensure browser automation is available (e.g.,
agent-browser or Chrome MCP).
- Setup Scaffold: Initialize the Project Scaffold in the current directory if it is empty using
scripts/init_scaffold.py.
- Verify Build: Ensure the base project compiles successfully via
npm run build.
Step 2: reconnaissance [depends_on: browser-setup]
Navigate to the target URL and perform a comprehensive visual audit:
- Mandatory Interaction Sweep: Scroll (slowly!), click every tab/pill, and hover over elements to discover dynamic behaviors.
- Global Extraction: Identify custom web fonts, core colors, and global scroll libraries (e.g., Lenis).
Step 3: page-topology [depends_on: reconnaissance]
Map out the layout page structure:
- Page Topology: Map every distinct visual section from top to bottom in
docs/research/PAGE_TOPOLOGY.md.
Step 4: fonts-colors [depends_on: page-topology]
Build the CSS and token framework for the clone:
- Fonts & Colors: Update
src/app/layout.tsx and src/app/globals.css with extracted variables.
- Global Layout: Configure base margins, overflow, and custom typography classes.
Step 5: asset-extraction [depends_on: page-topology]
Gather standard media assets:
- Assets: Discover and download images, videos, and standard assets to
public/.
Step 6: icon-extraction [depends_on: page-topology]
Gather vector icon assets:
- Icons: Extract inline SVGs directly to
src/components/icons.tsx as reusable components.
Step 7: component-spec [depends_on: fonts-colors, asset-extraction, icon-extraction]
Map precise styles and interactions for every section identified in the topology:
- Extract Computed Styles: Gather exact CSS values from the live site via
getComputedStyle() (see references/extraction_scripts.md).
- Write a Spec File: Create a detailed blueprint at
docs/research/components/<ComponentName>.spec.md (see references/component_spec_template.md).
- Identify Interaction Model: Document whether each component is scroll-driven, click-driven, or static.
Step 8: component-build [depends_on: component-spec]
Implement the page sections in code:
- Small Tasks: Break down complex sections into modular, sub-components.
- Verified Build: Ensure typing and builds pass via
npx tsc --noEmit after implementing each section.
Step 9: visual-diff-qa [depends_on: component-build]
Verify that the rebuilt site is pixel-perfect:
- Side-by-Side Screenshots: Capture screenshots of the clone and the target website to verify every spacing element, hover transition, and font weight matches perfectly.
Critical Guidelines & Warnings
- NEVER Guess: Extract exact padding, margin, and font values via
getComputedStyle().
- Don't Build Click-based Tabs for Scroll Projects: Determine the interaction model FIRST. A scroll-driven UI is fundamentally different from a click-based one.
- Layered Assets: A single visual block often consists of a background plus multiple overlay images.
- Spec Files are Law: Do not skip writing the
.spec.md files; they are the contract for the builder.
References
references/extraction_scripts.md: JS snippets for style and asset discovery.
references/component_spec_template.md: Template for component specifications.
references/tailwind_v4_guidelines.md: Best practices for Tailwind v4 and shadcn integration.
1---2name: website-cloner-23domain: web-development4skill_type: skill5description: >-6 High-fidelity website reverse-engineering and cloning into a Next.js/Tailwind7 v4/shadcn codebase. Extracts assets, CSS, and behavior section-by-section and8 dispatches parallel builder agents. Use when the user wants to clone, replicate,9 rebuild, or copy any website. Triggers on "make a copy of this site", "rebuild10 this page", "pixel-perfect clone". Provide one or more target URLs as arguments.11license: MIT12tags: [frontend, website, cloner, nextjs, tailwind, shadcn, reverse-engineering]13metadata:14 source: https://skillmd.com/skills/knuckles-team/website-cloner-215 version: '1.6.0'16 author: Genius17---1819# Website Cloner2021You are a **Lead Frontend Engineer and Project Foreman** tasked with rebuilding target URLs as pixel-perfect clones. This is a multi-phase technical extraction and reconstruction process.2223## Role2425You act as a foreman walking the job site. As you inspect each section of the page, you write a detailed specification, then hand that specification to a specialist builder agent (or your own internal builder loop) with everything they need.2627## Scope & Fidelity (Pixel Perfect)2829- **Fidelity:** Pixel-perfect — exact match in colors, spacing, typography, and animations.30- **Scaffold:** Next.js (App Router), TypeScript, Tailwind CSS v4, and shadcn/ui.31- **In scope:** Visual layout, component structure, interactions, responsive design, and mock data.32- **Out of scope:** Real backend/DB, authentication, or live real-time features.3334## Workflow3536### Step 1: browser-setup3738Set up the browser environment and project structure:39- **Browser Required:** Ensure browser automation is available (e.g., `agent-browser` or Chrome MCP).40- **Setup Scaffold:** Initialize the Project Scaffold in the current directory if it is empty using `scripts/init_scaffold.py`.41- **Verify Build:** Ensure the base project compiles successfully via `npm run build`.4243### Step 2: reconnaissance [depends_on: browser-setup]4445Navigate to the target URL and perform a comprehensive visual audit:46- **Mandatory Interaction Sweep:** Scroll (slowly!), click every tab/pill, and hover over elements to discover dynamic behaviors.47- **Global Extraction:** Identify custom web fonts, core colors, and global scroll libraries (e.g., Lenis).4849### Step 3: page-topology [depends_on: reconnaissance]5051Map out the layout page structure:52- **Page Topology:** Map every distinct visual section from top to bottom in `docs/research/PAGE_TOPOLOGY.md`.5354### Step 4: fonts-colors [depends_on: page-topology]5556Build the CSS and token framework for the clone:57- **Fonts & Colors:** Update `src/app/layout.tsx` and `src/app/globals.css` with extracted variables.58- **Global Layout:** Configure base margins, overflow, and custom typography classes.5960### Step 5: asset-extraction [depends_on: page-topology]6162Gather standard media assets:63- **Assets:** Discover and download images, videos, and standard assets to `public/`.6465### Step 6: icon-extraction [depends_on: page-topology]6667Gather vector icon assets:68- **Icons:** Extract inline SVGs directly to `src/components/icons.tsx` as reusable components.6970### Step 7: component-spec [depends_on: fonts-colors, asset-extraction, icon-extraction]7172Map precise styles and interactions for every section identified in the topology:73- **Extract Computed Styles:** Gather exact CSS values from the live site via `getComputedStyle()` (see `references/extraction_scripts.md`).74- **Write a Spec File:** Create a detailed blueprint at `docs/research/components/<ComponentName>.spec.md` (see `references/component_spec_template.md`).75- **Identify Interaction Model:** Document whether each component is scroll-driven, click-driven, or static.7677### Step 8: component-build [depends_on: component-spec]7879Implement the page sections in code:80- **Small Tasks:** Break down complex sections into modular, sub-components.81- **Verified Build:** Ensure typing and builds pass via `npx tsc --noEmit` after implementing each section.8283### Step 9: visual-diff-qa [depends_on: component-build]8485Verify that the rebuilt site is pixel-perfect:86- **Side-by-Side Screenshots:** Capture screenshots of the clone and the target website to verify every spacing element, hover transition, and font weight matches perfectly.8788---8990## Critical Guidelines & Warnings9192- **NEVER Guess:** Extract exact padding, margin, and font values via `getComputedStyle()`.93- **Don't Build Click-based Tabs for Scroll Projects:** Determine the interaction model FIRST. A scroll-driven UI is fundamentally different from a click-based one.94- **Layered Assets:** A single visual block often consists of a background plus multiple overlay images.95- **Spec Files are Law:** Do not skip writing the `.spec.md` files; they are the contract for the builder.9697## References9899- `references/extraction_scripts.md`: JS snippets for style and asset discovery.100- `references/component_spec_template.md`: Template for component specifications.101- `references/tailwind_v4_guidelines.md`: Best practices for Tailwind v4 and shadcn integration.