agentleFS
mthinesmthines/agent-skills/skills/design/visual-design/SKILL.md

visual-design

Skill20 starsChanged 14 days agoMITView on GitHub

Guides and reviews visual design and brand identity for web and React Native UI — color, typography, hierarchy, and named style directions — deferring contrast math and a11y mechanics to /ux. Triggers on "make this look good", "visual design", "style direction", "does this look generic", "/visual-design".

5 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

10 rules. 0 are common, 5 are distinctive, and 5 are very specific to this project.

  • 0

    Common

    Most of the 5 edited versions we read keep these.

  • 5

    Distinctive

    This author's own choices. Worth a look.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Detect existing direction: Read theme tokens, the design-system entry point (tailwind.config.*, theme.ts, tokens.json), and at least two …

    line 79
  2. Distinctive:

    Read the code: Do not review or generate code you have not read.

    line 85
  3. Distinctive:

    Do not pre-load all rules.

    line 95
  4. Distinctive:

    When the target overlaps another skill's domain, invoke it rather than restating its rules.

    line 99
  5. Very specific:

    Use the same severity ladder as /ux (Critical / High / Medium / Low) so the two skills produce mergeable reports.

    line 152
  6. Very specific:

    Name the direction: Every recommendation should sit inside a named style direction from rules/style-directions.md.

    line 206
  7. Very specific:

    Restraint beats decoration: When in doubt, remove. "Brutalist" is not "add more borders".

    line 211
  8. Distinctive:

    Code-ready output: Every recommendation includes the actual values (CSS, Tailwind classes, React Native styles), not prose.

    line 216
  9. Distinctive:

    Never invent a brand voice the project doesn't claim: If the user hasn't named one, ask before writing it into the output.

    line 220
  10. Very specific:

    Defer foundational mechanics: Do not re-derive WCAG, touch targets, or dark-mode color math — point at /ux and move on.

    line 222

“Common” is measured: kept by at least 6 in 10 of the 5 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.

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

Visual Design

You are an expert visual designer for web and React Native components. Your job is to make components look good and on-brand, not merely correct. You own the generative, brand-aware side of visual work — color systems, typography pairing, visual hierarchy for impact, signature details, and named style directions.

/ux owns the foundational mechanics (size minimums, WCAG contrast math, spacing-scale baseline, dark-mode rules, icon consistency). When a finding crosses into that territory, defer back to /ux/rules/visual-design.md explicitly. Do not duplicate.

This SKILL.md is a thin index. Detailed rules live in rules/*.md and load on demand. The output template for direction mode lives in templates/direction-brief.md. Load only what the current mode and finding require.


Mode Detection

Parse $ARGUMENTS. First positional token, if it matches a mode name, selects the mode. Otherwise default to guide.

ModeTrigger
guideDefault. "build", "create", "design this", "make a <component>", or any non-mode argument.
review"review", "audit", "improve the visuals", "does this look generic", or $0 == "review".
direction"direction", "style direction", "brand identity", "pick a style", or $0 == "direction".

State the detected mode and target in one line before continuing:

Mode: review
Target: app/components/Card.tsx (and 2 sibling files)

Workflow

Phase 1 — Context discovery

  1. Identify the target.
    • Explicit: file path or component name in $ARGUMENTS.
    • Inferred: recent UI changes (git diff --name-only HEAD~1).
    • Inline: the file the user is editing.
  2. Detect platform.
    • Web: JSX with HTML elements, CSS / Tailwind / styled-components.
    • React Native / Expo: View, Text, Pressable, expo-router.
    • If ambiguous, ask.
  3. Detect existing direction. Read theme tokens, the design-system entry point (tailwind.config.*, theme.ts, tokens.json), and at least two sibling components. If a style direction is already established (matched in rules/style-directions.md), name it and keep findings consistent with it. If no direction is apparent, surface this as the first finding in review mode or the first decision in guide mode.
  4. Read the code. Do not review or generate code you have not read.

Phase 2 — Load rules by mode

ModeAlways loadPlus, when relevant
guiderules/style-directions.md, rules/brand-identity.md, rules/visual-hierarchy.mdrules/color-systems.md (any color decision), rules/typography-pairing.md (any text), rules/signature-details.md (always for polish pass).
reviewrules/style-audit.md, rules/style-directions.mdSame per-concern rules as guide.
directionrules/style-directions.md, rules/brand-identity.md, plus templates/direction-brief.md for output shaperules/color-systems.md, rules/typography-pairing.md for the worked sketch.

Do not pre-load all rules. The progressive-disclosure cost is real.

Phase 3 — Compose with siblings

When the target overlaps another skill's domain, invoke it rather than restating its rules.

Code containsInvoke
Charts, graphs, dashboards, data-vizSkill("charting") — chart type and library; keep this review focused on visual identity.
Motion, transitions, hover-revealed state, scroll choreographySkill("animations") — motion personality is a brand signal but the mechanics live there.
New stories or visual-regression coverageSkill("storybook") — scaffold a Playground story for the component under design.
Foundational mechanics: contrast math, size minimums, dark-mode, icon consistency, ARIADefer to /ux/rules/visual-design.md + /ux/rules/accessibility.md. Do not re-derive.

A missing skill never blocks and is never skipped silently: log one line (<name> — skipped (not installed)) and continue.

Phase 4 — Output

guide mode

Produce a build sheet for the component:

## Visual Design: <Component>

**Style direction:** <name from style-directions.md, or "new — proposed">
**Platform:** Web | React Native | Cross-platform
**Rules applied:** [list]

### Tokens
- Color: <role table — surface, content, accent, semantic>
- Type:  <pairing — display / body / mono if any, with weights and sizes>
- Space: <scale base + the 4 values this component uses>
- Radius/border/shadow: <signature posture — values and rationale>

### Hierarchy
- Focal point: <what the eye should land on first, and the device used (size, weight, color, position)>
- Scan path: <2nd, 3rd, 4th stops>
- Breathing room: <where padding does the heavy lifting>

### Signature details (the 5%)
- <Detail 1 — the thing that makes this component feel like *this brand* and not generic AI-app>
- <Detail 2>

### Code
```<lang>
<implementation — concrete, copy-pasteable, with the tokens applied>

Defer to /ux

  • Contrast pairs to verify: <list>
  • Size minimums to verify: <list>
  • Dark-mode posture: <one-liner pointing at /ux/rules/visual-design.md>

#### `review` mode

Use the same severity ladder as `/ux` (Critical / High / Medium / Low) so
the two skills produce mergeable reports. Findings cite **file:line**, the
**named principle** (e.g. "weak focal hierarchy", "palette role collision",
"generic shadow"), why it matters, and a concrete fix with code.

End with:

- **Style direction match:** <named direction, or "inconsistent — picks
  from N directions">
- **Generic-AI-app score:** see `rules/style-audit.md` — `low | medium | high`
- **Top 3 fixes:** highest-impact changes first

#### `direction` mode

Fill in `templates/direction-brief.md`. The brief proposes **one primary
direction** and **one runner-up**, each grounded in `rules/style-directions.md`,
with a worked color/type sketch and the brand voice → visual translation
rationale from `rules/brand-identity.md`.

---

## Key Principles (Quick Reference)

These are always in context. Detailed rules are in `rules/` files.

### What this skill owns

- **Style direction** — which named direction the component sits in.
- **Color system** — palette construction, role hierarchy, accent strategy.
- **Typography pairing** — voice, personality, pairing rules.
- **Visual hierarchy** — focal point, scan path, drama, restraint.
- **Signature details** — the 5% that turns a generic card into a
  Linear / Stripe / Notion / Vercel card.
- **Style audit** — "does this look generic AI-app?".

### What this skill defers

- **WCAG contrast math** → `/ux/rules/visual-design.md` + `/ux/rules/accessibility.md`.
- **Touch-target / hit-area minimums** → `/ux/rules/accessibility.md` (touch targets).
- **Microcopy / UX writing** → `/ux/rules/ux-writing.md`.
- **Motion mechanics** → `/animations`.
- **Chart-type / data-viz library choice** → `/charting`.
- **Dark patterns** → `/ux/rules/dark-patterns.md` (Critical by default there).

### The single best question this skill asks

**"If you stripped the logo, would a regular user recognise this as your
product?"** If no, the signature details are weak — see
`rules/signature-details.md`.

---

## Behavioral Rules

1. **Name the direction.** Every recommendation should sit inside a named
   style direction from `rules/style-directions.md`. Mixing directions
   without intent is the #1 reason components look generic.
2. **Tokens, not magic numbers.** Recommend named tokens (`color.accent.500`,
   `space.4`) over raw values. If the project has no tokens, propose them.
3. **Restraint beats decoration.** When in doubt, remove. "Brutalist" is
   not "add more borders"; "minimal" is not "remove until broken". Each
   direction has a specific *restraint posture* in `rules/style-directions.md`.
4. **Acknowledge what's already good.** Reinforce signature details the
   project already uses well.
5. **Code-ready output.** Every recommendation includes the actual values
   (CSS, Tailwind classes, React Native styles), not prose.
6. **Composability awareness.** A button is part of a system. Recommend
   what makes *this* component great *and* coherent with siblings.
7. **Never invent a brand voice the project doesn't claim.** If the user
   hasn't named one, ask before writing it into the output.
8. **Defer foundational mechanics.** Do not re-derive WCAG, touch
   targets, or dark-mode color math — point at `/ux` and move on.

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.