replica-design
Skill399 starsChanged 40 days agoMITView on GitHub
Rebuilds an app's design system for a clone: colour roles, type scale, spacing, radius, shadows and every component with its states, as design tokens plus component specs, with original assets instead of the target's logos, icons, illustrations or licensed fonts. Includes a WCAG contrast checker. Use when the user says "match the design", "rebuild the design system", "get the colours and fonts", "make it look like X", "design tokens for my clone", or after /replica-architect.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- realtahailyas/reverse-engineer-app-skillsword for word, first seen Oct 25, 2026
Other versions1 project uses a version of this.
Word for word · 1
- realtahailyas/reverse-engineer-app-skills★ 3100% same
Edited · 0
None found.
Wrote their own version · 0
None found.
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 repository10 other files it gives its agents.
- replica-architectSkill
- replica-backendSkill
- replica-brandSkill
- replica-buildSkill
- replica-deploySkill
- replica-diffSkill
- replica-entrepreneurSkill
- replica-launchSkill
- replica-reconSkill
- replica-testSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 4 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
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 29
Logos, icons, illustrations, photos, sounds: Use an open icon set (Lucide, Phosphor, Heroicons, Tabler, all MIT or similar) and make or …
- Distinctive: line 32
Licensed fonts: If the original uses a paid or proprietary font, pick an open one with the same job: Inter, Geist, IBM Plex, Manrope, …
- Distinctive: line 34
Their copy: Every label and empty state gets written fresh.
- Very specific: line 59
Add a pairs list for every text and background combination the app uses, then:
- Distinctive: line 84
Every state the recon saw, plus the ones it should have: focus, disabled, loading, error, empty.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
replica-design
Reads replica/recon.md and the screenshots in replica/screens/. Writes
replica/design/tokens.json (template in this folder), tokens.css, the
Tailwind mapping, and replica/design/components.md.
python3 contrast.py replica/design/tokens.json # every text pair, WCAG ratioThe rules
What you rebuild is the system: the roles, the scale, the patterns, the way a form or a modal behaves. Those are not ownable, and users expect them. What you never take:
- Logos, icons, illustrations, photos, sounds. Use an open icon set (Lucide, Phosphor, Heroicons, Tabler, all MIT or similar) and make or commission your own illustrations. Do not trace theirs.
- Licensed fonts. If the original uses a paid or proprietary font, pick an open one with the same job: Inter, Geist, IBM Plex, Manrope, Source Serif.
- Their copy. Every label and empty state gets written fresh.
- Their brand colour. Record it as a role (
accent), use a neutral placeholder now, and replica-brand gives you your own. The signature colour plus the signature layout is trade dress, and it changes before launch.
Step 1: measure, do not guess
From the screenshots (zoom in, use a colour picker on the user's machine):
- Colour roles, not colours: bg, surface, border, border-input, text, text-muted, accent, on-accent, danger, success, warning. Count how many greys the app really uses. Usually 5 to 7.
- Type scale: sizes, line heights, weights. Snap to a scale (12, 14, 16, 20, 28, 40 is common). Note the font category, not the font file.
- Spacing: measure gaps between elements. It is almost always a 4 or 8 base. Write the scale.
- Radius, shadow, motion: two or three of each.
- Layout: max content width, grid, breakpoints, sidebar width, header height.
Step 2: write the tokens
Fill tokens.json. Keep the role names. replica-brand only changes values.
Generate tokens.css as custom properties and map them into Tailwind's theme
so components use bg-surface text-muted, never raw hex.
Add a pairs list for every text and background combination the app uses,
then:
python3 contrast.py replica/design/tokens.jsonAA is the floor: 4.5:1 for body text, 3:1 for large text and for input borders and focus rings. It exits 1 on a failure. Fix it in the tokens, not per component.
Step 3: component specs
For every component in the recon list, one block in components.md:
Button
variants primary, secondary, ghost, danger
sizes sm 32px, md 40px, lg 48px
states default, hover, active, focus-visible (2px ring, accent), disabled, loading
tokens bg accent, text on-accent, radius md, font sm/600
a11y real <button>, visible focus, loading keeps the label for screen readers
used on S02, S07, S09Every state the recon saw, plus the ones it should have: focus, disabled, loading, error, empty. Keyboard and screen reader behaviour is part of the spec.
Step 4: build the primitives
Build the components in code once, in isolation (a /design route or
Storybook), before any screen. Use an accessible base if the stack has one
(Radix, shadcn/ui, React Aria). Screenshot the page. That is the design system
check.
Output
tokens.json, tokens.css, the Tailwind config, components.md, the
primitives built, and a contrast report with zero AA failures. Next:
/replica-build.
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.

