Tokens logo

Tokens

Community
dawitlabs
tokens

Design token audit and enforcement skill. Finds every hardcoded color, spacing value, shadow, border radius, font size, and z-index in the codebase, maps them to an existing or proposed token system, then replaces hardcoded values with tokens — making the design system maintainable, themeable, and dark-mode-ready. Use when user wants to set up a token system, enforce design consistency, prepare for dark mode, or invokes /tokens.

Overview

Publisherdawitlabs
Repositoryui-skills
Skill nametokens
Stars
7
Forks
0
Bundled files
1
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 1 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by dawitlabs on GitHub. Read the source before you install it.

Installation

Install the Tokens AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/dawitlabs/ui-skills/tree/master/skills/tokens
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/dawitlabs/ui-skills.git /tmp/ui-skills
mkdir -p .claude/skills
cp -r /tmp/ui-skills/skills/tokens .claude/skills/tokens
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Tokens in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Tokens on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Tokens is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

/tokens — Design Token System

Audits every hardcoded design value in the codebase, builds or enforces a three-layer token system (primitives → semantic → component), and replaces hardcoded values with tokens throughout.

Token architecture and naming conventions: SYSTEMS.md

If a DESIGN-BRIEF.md exists in the project root: read it before starting. The brief's Color System and Typography fields give you the exact starting point for primitive tokens — use those hex values and font names directly rather than deriving them from scratch.


Phase 0 — Environment Setup

0a. Find source root

bash
find . -type d -name "src" \
  -not -path "*/node_modules/*" \
  -not -path "*/.next/*" \
  -not -path "*/dist/*" \
  -not -path "*/.turbo/*" \
  | head -1

If no src/: find . -type d -name "app" -not -path "*/node_modules/*" -not -path "*/.next/*" | head -1

Use the result as <SRC> throughout.

0b. Detect CSS-in-JS

bash
grep -r "styled-components\|@emotion\|@stitches\|vanilla-extract\|linaria" package.json packages/*/package.json apps/*/package.json 2>/dev/null | head -5

If found, the color/spacing greps in Phase 1 must also scan .ts files and template literals — the Phase 1 commands include these where needed. Note which library is in use.

0c. Detect the existing token system

Run all of these — note which ones return results:

bash
# CSS custom properties (most common)
find . -name "globals.css" -o -name "variables.css" -o -name "tokens.css" | grep -v node_modules

# Tailwind config
find . -name "tailwind.config.*" -not -path "*/node_modules/*"

# Style Dictionary / tokens.json
find . -name "tokens.json" -o -name "design-tokens.json" | grep -v node_modules
find . -type d -name "tokens" -not -path "*/node_modules/*"

# shadcn/ui pattern (CSS vars in globals.css under :root)
grep -rn ":root" . --include="*.css" -l | grep -v node_modules | head -5

# Existing token usage (CSS vars already in use)
grep -rn "var(--" "$SRC" --include="*.tsx" --include="*.css" | head -10

Map the result to one of these token homes — all fixes go here:

  • CSS custom properties → globals.css :root block
  • Tailwind theme.extend → tailwind.config.ts
  • Style Dictionary → tokens/ directory JSON files
  • None → create globals.css with :root block (see SYSTEMS.md)

0d. Snapshot the current token set

Read the token home file(s) found in 0b. List every existing token name and value. This is the baseline — new tokens must not conflict with or duplicate these.

0e. Check / start dev server (for visual verification)

bash
curl -s --max-time 2 -o /dev/null -w "%{http_code}" http://localhost:3000

If not 200, start the dev command and wait:

bash
until curl -s --max-time 2 http://localhost:3000 > /dev/null 2>&1; do sleep 1; done && echo "ready"

0f. Write the screenshot script

Write to the project root (needs node_modules). Deleted in Phase 4.

_tokens_screenshot.mjs:

js
import { chromium } from '@playwright/test';
import { mkdirSync } from 'fs';

const get = (f) => { const i = process.argv.indexOf(f); return i !== -1 ? process.argv[i+1] : null; };
const url    = get('--url');
const output = get('--output') ?? '/tmp/tokens-screenshot.png';
const width  = parseInt(get('--width')  ?? '1280');
const height = parseInt(get('--height') ?? '900');
const auth   = get('--auth');

if (!url) { console.error('--url required'); process.exit(1); }
mkdirSync('/tmp/tokens', { recursive: true });

const browser = await chromium.launch();
const context = await browser.newContext({
  viewport: { width, height },
  ...(auth ? { storageState: auth } : {}),
});
const page = await context.newPage();
try {
  await page.goto(url, { waitUntil: 'networkidle', timeout: 30000 });
} catch {
  await page.goto(url, { waitUntil: 'domcontentloaded', timeout: 30000 });
}
await page.screenshot({ path: output, fullPage: true });
await browser.close();
console.log('saved:', output);

After every screenshot: use the Read tool on the image.

0g. Handle auth-protected screens

bash
ls /tmp/uiux-auth-state.json /tmp/animate-auth-state.json /tmp/copy-auth-state.json /tmp/a11y-auth-state.json 2>/dev/null | head -1

If found, pass --auth <path> to all screenshot calls for protected routes.


Phase 1 — Token Audit

1a. Find all hardcoded colors

bash
SRC=<your source root>

# Hex colors in TSX/JSX (inline styles, arbitrary Tailwind values)
echo "=== Hex colors in TSX ===" && \
grep -rn "#[0-9a-fA-F]\{3,8\}" "$SRC" --include="*.tsx" --include="*.jsx" \
  | grep -v "//.*#[0-9a-fA-F]" | grep -v "\.md"

# Hex colors in CSS
echo "=== Hex colors in CSS ===" && \
grep -rn "#[0-9a-fA-F]\{3,8\}" "$SRC" --include="*.css" --include="*.scss" \
  | grep -v "^.*:.*var(--"

# rgb/hsl inline values
echo "=== RGB/HSL values ===" && \
grep -rn "rgba\?\|hsla\?" "$SRC" --include="*.tsx" --include="*.css" | grep -v "var(--"

# Arbitrary Tailwind color values (hex)
echo "=== Arbitrary Tailwind colors (hex) ===" && \
grep -rn "\(bg\|text\|border\|ring\|shadow\|fill\|stroke\)-\[#" "$SRC" --include="*.tsx"

# Named Tailwind palette colors that should be semantic tokens
# e.g. bg-blue-500 should become bg-primary if blue-500 is the primary color
echo "=== Named Tailwind palette colors ===" && \
grep -rn "\(bg\|text\|border\|ring\|fill\|stroke\)-\(slate\|gray\|zinc\|neutral\|stone\|red\|orange\|amber\|yellow\|lime\|green\|emerald\|teal\|cyan\|sky\|blue\|indigo\|violet\|purple\|fuchsia\|pink\|rose\)-[0-9]\{2,3\}" "$SRC" --include="*.tsx" \
  | grep -v "hover:\|focus:\|dark:\|disabled:" | head -40
# Note: filter hover/focus/dark variants — those are less likely to be primary brand usage
# High-frequency named colors (>5 occurrences) are strong tokenization candidates

If CSS-in-JS detected (styled-components / emotion): also run:

bash
# Hex colors in template literals
grep -rn "#[0-9a-fA-F]\{3,8\}" "$SRC" --include="*.ts" --include="*.tsx" \
  | grep "styled\.\|css\`\|createGlobalStyle\|keyframes" | grep -v "//.*#"

# CSS property values inside template literals
grep -rn "color:\s*#\|background:\s*#\|border.*:\s*#" "$SRC" --include="*.ts" --include="*.tsx"

1b. Find all hardcoded spacing

bash
# Off-grid pixel values in CSS (not on 4px scale)
echo "=== Off-grid spacing in CSS ===" && \
grep -rn "\b[0-9]*[13579]\b\s*px\|[0-9]\+\.\s*[0-9]\+\s*px" "$SRC" --include="*.css" \
  | grep -v "border\|outline\|shadow\|translate"

# Arbitrary Tailwind spacing values
echo "=== Arbitrary Tailwind spacing ===" && \
grep -rn "\(p\|m\|gap\|space\|w\|h\|inset\|top\|left\|right\|bottom\)-\[[0-9]" "$SRC" --include="*.tsx"

# Inline style pixel values
echo "=== Inline style px values ===" && \
grep -rn "style={{" "$SRC" --include="*.tsx" | grep "[0-9]px\|[0-9]rem" | head -20

1c. Find all hardcoded typography values

bash
# Font sizes in CSS not from scale
echo "=== Hardcoded font sizes ===" && \
grep -rn "font-size:" "$SRC" --include="*.css" | grep -v "var(--"

# Arbitrary Tailwind font sizes
echo "=== Arbitrary Tailwind font sizes ===" && \
grep -rn "text-\[[0-9]" "$SRC" --include="*.tsx"

# Font weights not from scale
echo "=== Hardcoded font weights ===" && \
grep -rn "font-weight:\s*[0-9]" "$SRC" --include="*.css" | grep -v "var(--"

# Line heights
echo "=== Hardcoded line heights ===" && \
grep -rn "line-height:\s*[0-9]" "$SRC" --include="*.css" | grep -v "var(--"

1d. Find all hardcoded shadows, radii, and z-indices

bash
# Box shadows not from token
echo "=== Hardcoded shadows ===" && \
grep -rn "box-shadow:" "$SRC" --include="*.css" | grep -v "var(--" && \
grep -rn "shadow-\[" "$SRC" --include="*.tsx"

# Border radii not from token
echo "=== Hardcoded border radii ===" && \
grep -rn "border-radius:" "$SRC" --include="*.css" | grep -v "var(--" && \
grep -rn "rounded-\[[0-9]" "$SRC" --include="*.tsx"

# Magic z-index numbers
echo "=== Magic z-index ===" && \
grep -rn "z-index:\s*[0-9]\|z-\[[0-9]" "$SRC" --include="*.css" --include="*.tsx"

1e. Find duplicate and inconsistent values

bash
# Same color used with different names (group by hex value to find duplicates)
grep -rh "#[0-9a-fA-F]\{3,8\}" "$SRC" --include="*.tsx" --include="*.css" \
  | grep -oE "#[0-9a-fA-F]{3,8}" | sort | uniq -c | sort -rn | head -20

High-count values = the most-used hardcoded colors. These are highest priority for tokenization.

1f. Screenshot every major screen (before state)

Take desktop and mobile for each screen — token changes to spacing and typography can break mobile layouts that look fine at 1280px.

bash
node _tokens_screenshot.mjs --url http://localhost:3000 --output /tmp/tokens/home-before-desktop.png
node _tokens_screenshot.mjs --url http://localhost:3000 --output /tmp/tokens/home-before-mobile.png --width 375 --height 812

Take 3–5 of the most visually distinct screens. Read every image. These are your before-state baseline.

Present the full audit findings grouped by category before continuing.


Phase 2 — Token Map

Before writing any code, build a token map — a translation table from every hardcoded value to its proposed token name.

Format:

Hardcoded value | Occurrences | Proposed token name | Layer
#3b82f6         | 23          | --color-primary      | semantic
#1d4ed8         | 8           | --color-primary-dark  | semantic
#f9fafb         | 31          | --color-surface       | semantic
13px            | 4           | --spacing-3-5         | primitive

Rules for naming (from SYSTEMS.md):

  • Primitive tokens: --color-{hue}-{shade}, --space-{n}, --radius-{size}
  • Semantic tokens: --color-primary, --color-surface, --color-text, --color-border
  • Component tokens: --button-bg, --card-shadow (only when component-specific)
  • Never name by visual appearance: --blue or --dark-gray → use semantic name
  • Semantic names survive color changes: --color-primary stays right when brand changes from blue to green

Group by token layer:

  1. New primitive tokens — raw values not already in the system
  2. New semantic tokens — meaning-based aliases to primitives
  3. Existing tokens to use — values that map to already-defined tokens

Ask user to confirm the token map before starting Phase 3.


Phase 3 — Recursive Token Loop

Work category by category: colors → spacing → typography → shadows → radii → z-index. Maximum 3 iterations per category.

3a. Define new tokens

Add all new primitive and semantic tokens to the token home file before replacing any usage.

CSS custom properties pattern:

css
/* In globals.css */
:root {
  /* Primitives */
  --color-blue-500: #3b82f6;
  --color-blue-700: #1d4ed8;
  --space-1: 4px;
  --space-2: 8px;

  /* Semantic */
  --color-primary:      var(--color-blue-500);
  --color-primary-dark: var(--color-blue-700);
  --color-surface:      #f9fafb;
  --color-text:         #111827;
  --color-border:       #e5e7eb;
}

/* Dark mode — same semantic names, different values */
.dark {
  --color-primary:  var(--color-blue-400);
  --color-surface:  #111827;
  --color-text:     #f9fafb;
  --color-border:   #374151;
}

Tailwind pattern:

ts
// tailwind.config.ts
export default {
  theme: {
    extend: {
      colors: {
        primary:  'var(--color-primary)',
        surface:  'var(--color-surface)',
        border:   'var(--color-border)',
      },
      spacing: {
        '3.5': 'var(--space-3-5)',
      },
    },
  },
}

3b. Replace hardcoded values in components

Work file by file through the highest-occurrence values first. For each:

  1. Read the file
  2. Replace the hardcoded value with the token
  3. Never change logic, layout, or component structure — only the value

CSS replacement:

css
/* Before */
color: #3b82f6;

/* After */
color: var(--color-primary);

Tailwind class replacement:

tsx
/* Before */
<div className="bg-[#3b82f6] text-[#111827]">

/* After */
<div className="bg-primary text-text">

Inline style replacement (last resort — prefer moving to CSS/Tailwind):

tsx
/* Before */
style={{ backgroundColor: '#3b82f6' }}

/* After — move to CSS class if possible */
className="bg-primary"

/* Or if dynamic: */
style={{ backgroundColor: 'var(--color-primary)' }}

3c. Screenshot after each category

bash
sleep 3  # hot reload
node _tokens_screenshot.mjs --url http://localhost:3000 --output /tmp/tokens/home-after-<category>-desktop.png
node _tokens_screenshot.mjs --url http://localhost:3000 --output /tmp/tokens/home-after-<category>-mobile.png --width 375 --height 812

Read both images. Compare against the before screenshots at both viewports.

The UI should look identical at both sizes. If anything changed:

  • Wrong color → token value is wrong, fix the token definition
  • Layout shift on mobile → a spacing/size token resolves to a different value than the hardcoded one, fix the token value

3d. Token checklist

Before advancing to the next category:

  • All values in this category replaced with tokens
  • No new hardcoded values introduced during replacement
  • Before/after desktop screenshots are visually identical
  • Before/after mobile screenshots (375px) are visually identical — spacing/typography tokens don't shift layout
  • Token names follow the primitive → semantic hierarchy
  • Dark mode block defines overrides for all semantic tokens (if dark mode exists)
  • No two tokens have the same value and different semantic meaning (consolidate duplicates)
  • No token name describes its appearance (--blue) — only its role (--color-primary)

Any fail → go back to 3b. After iteration 3: flag and advance.


Phase 4 — Cleanup & Report

Delete temp scripts:

bash
rm -f _tokens_screenshot.mjs

After each category:

✓ Colors (iteration 1) — 47 hardcoded values → 8 tokens
✓ Spacing (iteration 2) — 12 hardcoded values → 4 tokens

Final summary table:

CategoryHardcoded BeforeTokens CreatedValues ReplacedIterationsState
Colors............✓ / ⚠
Spacing............✓ / ⚠
Typography............✓ / ⚠
Shadows............✓ / ⚠
Radii............✓ / ⚠
Z-index............✓ / ⚠

Token inventory — all tokens defined, their values, and number of usages.

List all flagged hardcoded values that couldn't be resolved, with the reason.


Rules

  1. Define before you replace. Token must exist in the token home before any component references it.
  2. Primitive before semantic. Raw value first, semantic alias second, component token only when necessary.
  3. Name by role, never by appearance. --color-primary survives a rebrand. --color-blue doesn't.
  4. Screenshots are the regression test. Before/after must be visually identical. If they differ, the token value is wrong.
  5. Dark mode is the test for a good semantic layer. If swapping .dark changes colors correctly with no component edits, the tokens are right.
  6. One token per meaning. Two tokens with the same value and same meaning → merge them.
  7. Don't tokenize everything. border: 1px solid — the 1px doesn't need a token. Focus on values that change across themes or appear in multiple places.
  8. Max 3 iterations per category. Flag and move on.
  9. Never change logic. Only the design value — not the condition, component structure, or functionality.

See SYSTEMS.md for token architecture, naming conventions, and complete code patterns.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Tokens AI skill do?

Design token audit and enforcement skill. Finds every hardcoded color, spacing value, shadow, border radius, font size, and z-index in the codebase, maps them to an existing or proposed token system, then replaces hardcoded values with tokens — making the design system maintainable, themeable, and dark-mode-ready. Use when user wants to set up a token system, enforce design consistency, prepare for dark mode, or invokes /tokens.

Why use Tokens on TypingMind?

Because you install it once and use it with any model. Tokens is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Tokens in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/dawitlabs/ui-skills/tree/master/skills/tokens. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Tokens?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Tokens?

As many as you like. As long as a model supports skills, you can use Tokens with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Tokens AI skill free?

Yes. It is published on GitHub by dawitlabs under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇