clone-website
Skill194 starsChanged 4 months agoMITView on GitHub
Rebuild a live web page 1:1 from measurement into Figma, Penpot or code, then verify it. Use when reproducing a page or checking a rebuild against the original.
18 other projects use a version of this.
Adopted by
Most recently taken up by
- realbrianhanson/lovable-website-clonertheir own version, first seen Nov 15, 2026
- imMamdouhaboammar/Mimeratheir own version, first seen Nov 8, 2026
- coco-research/cocotheir own version, first seen Nov 1, 2026
- julianromli/ai-skillstheir own version, first seen Oct 25, 2026
- Shri-Phnx/claude-skillstheir own version, first seen Oct 13, 2026
- 444Nazky/ai-clonertheir own version, first seen Oct 11, 2026
Used with
Projects with it also use vercel-react-best-practices, frontend-design and 1 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
VercelSkillvercel-react-best-practices
React and Next.js performance optimization guidelines from Vercel Engineering. This skill should be used when writing, reviewing, or refactoring React/Next.js code to ensure optimal performance patterns. Triggers on tasks involving React components, Next.js pages, data fetching, bundle optimization, or performance improvements.
Also in 3 projects that use this
BlockSkillfrontend-design
Create distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, or applications. Generates creative, polished code that avoids generic AI aesthetics.
Also in 4 projects that use this
SentrySkillskill-creator
Create new agent skills following the Agent Skills specification. Use when asked to "create a skill", "add a new skill", "write a skill", "make a skill", "build a skill", or scaffold a new skill with SKILL.md. Guides through requirements, writing, registration, and verification.
Also in 3 projects that use this
Other versions18 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 18
- JCodesMore/ai-website-cloner-template★ 39ksame name
- yournextstore/yournextstore★ 5.5ksame name
- coco-research/coco★ 797same name
- julianromli/ai-skills★ 222same name
- Mahanaicoach/ai-site-cloner★ 31same name
- realbrianhanson/lovable-website-cloner★ 9same name
- david-vrba/claude-power-skills★ 4same name
- edorabma9304/Claude-Skill-by-Ozolab-Indonesia★ 2same name
- Shri-Phnx/claude-skills★ 2same name
- abuzar310/cloneforge★ 1same name
8 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository44 other files it gives its agents.
- algorithmic-color-paletteSkill
- app-shellSkill
- authentic-product-representationSkill
- brand-visual-languageSkill
- button-statesSkill
- color-mode-and-themeSkill
- component-family-consistencySkill
- coordinated-data-viewsSkill
- data-display-and-selectionSkill
- dembrandtSkill
- domain-expert-configurationSkill
- elevation-and-depthSkill
- extract-designSkill
- form-designSkill
- generate-ui-from-brandSkill
- gestalt-ui-organisationSkill
- global-toolbar-controlsSkill
- information-architectureSkill
- layout-paradigms-and-consistencySkill
- loading-states-and-perceived-performanceSkill
- marketing-vs-product-systemSkill
- micro-interactionsSkill
- modal-and-overlay-patternsSkill
- modular-scale-typographySkill
- motion-and-storytellingSkill
- nielsen-usability-heuristicsSkill
- notifications-and-recoverySkill
- operational-expert-tool-uiSkill
- performance-and-web-vitalsSkill
- real-world-metaphorsSkill
- repeated-component-alignmentSkill
- responsive-paradigmsSkill
- scroll-areasSkill
- semantic-html-and-seoSkill
- sizing-unitsSkill
- status-colors-and-errorsSkill
- sticky-and-fixed-elementsSkill
- surface-separationSkill
- tab-navigationSkill
- ui-context-and-scopeSkill
- ui-densitySkill
- user-flows-and-guided-pathsSkill
- visual-emphasis-and-hierarchySkill
- wcag-accessibilitySkill
The file
We split this skill into the rules it gives an agent
4 rules. 0 are common, 3 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 18 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 50
Only the rebuild step cares whether the target is Figma, Penpot or a codebase.
- Very specific: line 88
Read the tokens off the live page rather than off a picture — the type scale, palette and spacing are already declared.
- Distinctive: line 142
Every remaining error is local, so the instruction has to be local too: name the section, name what is wrong, name the source value.
- Distinctive: line 154
Import gone, staging assets gone.
“Common” is measured: kept by at least 6 in 10 of the 18 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Clone a Website, Accurately
One-click cloners return a lookalike: it resembles the page, and nothing in the process ever checked whether it matches it. This skill produces a different artifact — values traceable to the source, and a diff that says what is still wrong.
The one rule: if the page states a value, read it. If it does not, and only then, decide it. A rebuild done that way is finished when the diff is empty, which you can check, rather than when it looks about right, which nobody can.
Only the rebuild step cares whether the target is Figma, Penpot or a codebase. Everything else is the same job.
Decide before you capture
- Two widths are the design target, 390 mobile and 1440 desktop. Real devices open the same page at other ratios, so these are what you design to, not what you will see.
- A site that follows the light and dark preference renders two different pages. Capture both, or say which one you copied.
- The IP, locale, consent state and A/B branch reach the site before you do, so what you capture is one visitor's view, not the page. Check from a second location before trusting prices or copy.
1. Capture the rendered page
Drive a real browser and take four things at once: full-page screenshots at both widths, the rendered DOM, the computed styles of each section, and the asset URLs the page requested. On a client-rendered page this is the only capture that contains anything at all.
Rendered state is the specification, not the file the server sent.
2. Take the raw source as well
Markup carries text verbatim and names every asset URL without a screenshot in the loop.
wget -r -l 3 -k -p -E -nc https://example.com/pageHost-spanning is left out on purpose: CDN assets stay unfetched, their URLs are
in the markup, and they get downloaded directly at step 4. wget does not run
JavaScript, which is why it is the second capture and not the first.
3. Measure the design values
Read the tokens off the live page rather than off a picture — the type scale,
palette and spacing are already declared. Use [[extract-design]] for the
extraction; --design-md gives you a brief you can hand straight to the
rebuild step.
4. Take content and images from the capture
Parse headings, body copy, numbers, lists, links and the footer out of the DOM, then download every referenced asset. Never retype a string and never infer a value the source states. Nothing retyped means nothing drifts.
5. Build section by section
One section at a time — header, hero, feature rows, tables, media, testimonials, FAQ, footer — in auto layout or real layout containers, never pinned coordinates. Screenshot each section and compare it before moving on. A rebuild judged only at the end gets judged once, badly.
6. Compare three things, not one
| Check | Question |
| Visual | Does it look the same at each width? |
| Structural | Is it built from sane containers rather than absolute positions? |
| Semantic | Are the words, links, numbers and variants the ones the source states? |
The third is the one people drop, and it is the one that matters most. A perfect visual copy can carry wrong data, and it will pass every screenshot comparison you run against it.
Token-level verification is free: extract the original, extract the rebuild, compare the two token sets. The difference is your remaining work, stated as a list instead of a feeling.
7. Fix what the capture could not see
Some errors are structural blindness, not degradation, and none of them look like errors:
- A hero using the wrong image because the real one is a CSS background.
- Carousel or video items in the wrong order.
- Two cards whose values differ by variant, copied from each other.
Report what you could not resolve rather than guessing.
8. Fonts: stand in, then swap
Licensed foundry fonts usually cannot be loaded in an automated environment, and a visual lookalike is the wrong answer because it moves every line break you are about to verify. Build on a metric-compatible stand-in, then swap to the real face at the end with one find and replace.
9. Fix by targeted prompt, not by rerunning
Every remaining error is local, so the instruction has to be local too: name the section, name what is wrong, name the source value. Rerunning the whole rebuild to fix one card costs the hour again and moves the errors somewhere new.
Where the fix is a judgement call rather than a value, produce two or three versions of that one section and ask which is right — and ask what made it right. That sentence is what gets applied to the next section instead of a guess.
10. Throw the scaffolding away
Import gone, staging assets gone. What is left is the two frames, named layers, icons as vectors, not one absolutely positioned element. If the import is still in the file, the file is still a draft.
What this is for
Reproducing a page is research: teardowns, redesign baselines, migration references, and regression checks against your own site. The last is the most undervalued — the fastest way to learn whether production still matches its design system is to extract it and compare.
Shipping someone else's layout, copy or brand as your own is a different act, and the pipeline being fast does not make it a better idea. Respect robots.txt and terms of service.
Related
- [[extract-design]] — the measurement step, and the verification pass
- [[generate-ui-from-brand]] — when the goal is new UI in a brand, not a copy
- [[authentic-product-representation]] — keeping rebuilt content truthful
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

