agentleFS
OneWave-AIOneWave-AI/claude-skills/claude-design-system-architect/SKILL.md

claude-design-system-architect

Skill378 starsChanged 52 days agoMITView on GitHub

Generate a complete, premium design system from a brand brief or an existing site — color tokens, type scale, spacing/radius/elevation, motion language, and component specs — exported as Tailwind config, CSS variables, and a usage doc. Built to look editorial and human-crafted, not template-generated. Use when starting a new product/site, unifying an inconsistent UI, or codifying a brand into reusable tokens.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every value is a token, nothing arbitrary · contrast meets AA · dark mode + reduced-motion included · no purple / no emoji unless the brand …

    line 60

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Claude Design System Architect

A design system is the difference between a site that looks bespoke and one that looks like every other AI-generated landing page. This skill builds the token layer and component language that makes everything downstream look intentional — then exports it in the formats your stack actually uses.

Pairs with claude-landing-composer (which consumes these tokens to build pages) and claude-design-critic (which audits against them).

Step 1 — Establish the brand inputs

From a brief, an existing site (WebFetch / screenshots), or a few reference sites the user likes:

  • Personality — 3–5 adjectives (e.g. "warm, editorial, confident, calm"). These drive every later decision.
  • Audience & context — who uses it, on what device, in what mood.
  • Existing equity — logo, locked colors, fonts already in use.
  • Anti-references — what it must NOT look like.

Defaults & house rules

Unless the brand explicitly overrides:

  • No purple anywhere in the palette.
  • No emoji in UI — use an icon set (Lucide / Heroicons) and specify it as a token.
  • Lean toward warm / earth tones (sand, terracotta, clay) and consider a dark surface option.
  • Motion is part of the system, not an afterthought.

Step 2 — Build the tokens

Design each layer deliberately and document the why:

  • Color — a real ramp (50–900) per hue, semantic tokens (surface, text, muted, accent, border, success/warn/danger), and verified contrast (WCAG AA min for body text). Include a dark mode set. No arbitrary hexes scattered later — everything maps to a token.
  • Type — a font pairing with rationale, a modular scale (not random px), line-heights and tracking per size, and weights. Prefer characterful, legible pairings over the default system stack that screams "AI made this."
  • Space & layout — a spacing scale, container widths, grid, breakpoints. Generous, asymmetric whitespace reads premium; cramped uniform padding reads generated.
  • Radius / border / elevation — a small, consistent set. Shadows that match the brand's weight (soft for warm brands, crisp for technical ones).
  • Motion — duration + easing tokens, named patterns (entrance, emphasis, transition), and a reduced-motion stance. This is what makes it feel alive.

Step 3 — Specify components

For the core set (button, input, card, nav, section, badge, modal), specify variants, states (hover/focus/active/disabled), sizing, and accessibility notes — all expressed in the tokens above, never one-off values.

Step 4 — Export

Produce real files, not prose:

  • tailwind.config extension (or @theme for Tailwind v4) wiring every token.
  • tokens.css — CSS custom properties (light + dark).
  • design-system.md — the usage doc: palette with hexes + contrast, type scale, spacing, motion, component specs, and do/don't examples that call out the AI-tell patterns to avoid.

Step 5 — Hand off

End with a one-screen summary and the next move: claude-landing-composer to build pages on these tokens, or claude-design-critic to retrofit an existing UI onto them.

What separates premium from generated (bake this in)

  • Intentional, restrained palette — not 6 competing accent colors.
  • A real type scale and characterful fonts — not default sans at 3 sizes.
  • Asymmetric, generous whitespace — not uniform padding everywhere.
  • Motion with personality — not the same fade-up on every element.
  • One confident accent — not rainbow gradients (and never purple-by-default).

Guardrails recap

Every value is a token, nothing arbitrary · contrast meets AA · dark mode + reduced-motion included · no purple / no emoji unless the brand demands it · export real config files, not descriptions.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.