Uiux logo

Uiux

Community
dawitlabs
uiux

Master UI/UX redesign skill that audits every screen in the project, ranks them by priority, then redesigns each one recursively until pixel-perfect — applying color theory, typography, hierarchy, spacing, and accessibility principles automatically. Use when user wants to improve, redesign, or polish the UI/UX of an app, says "make the UI better", "redesign this", "pixel perfect", or invokes /uiux.

Overview

Publisherdawitlabs
Repositoryui-skills
Skill nameuiux
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 Uiux 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/uiux
  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/uiux .claude/skills/uiux
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Uiux 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 Uiux 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 Uiux 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.

/uiux — Master UI/UX Redesign

A systematic, recursive redesign loop. Works through every screen from highest to lowest priority, iterating each one until pixel-perfect based on what the rendered UI actually looks like, not what the code says.

Before starting: read ~/.claude/skills/ui-design-principles/SKILL.md — all color, typography, spacing, and hierarchy rules live there and apply automatically throughout.

If a DESIGN-BRIEF.md exists in the project root: read it before touching any file. It sets the visual direction (palette, typography, component style, animation level, defining screen). Design decisions in the brief override defaults — don't re-derive what's already decided.


Phase 0 — Environment Setup

0a. Check / start dev server

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

If the result is not 200, find the dev command (pnpm dev, npm run dev, etc.) and start it in the background. Then wait for it to be ready:

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

0b. Verify Playwright

bash
node -e "require('@playwright/test')" 2>/dev/null && echo "ready" || echo "missing"

If missing:

bash
pnpm add -D @playwright/test && npx playwright install chromium

If installation is impossible, ask the user to share screenshots after each fix. Do not make visual claims without real visual input.

0c. Write the screenshot script

Write this file to the project root (so it can resolve node_modules). It stays for the whole session and is deleted in Phase 4.

_uiux_screenshot.mjs (write to project root):

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

const get = (flag) => {
  const i = process.argv.indexOf(flag);
  return i !== -1 ? process.argv[i + 1] : null;
};

const url    = get('--url');
const output = get('--output');
const width  = parseInt(get('--width')  ?? '1280');
const height = parseInt(get('--height') ?? '900');
const auth   = get('--auth');

if (!url || !output) { console.error('--url and --output required'); process.exit(1); }

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

Usage — desktop:

bash
node _uiux_screenshot.mjs --url http://localhost:3000<ROUTE> --output /tmp/uiux-<name>-desktop.png

Usage — mobile (375px):

bash
node _uiux_screenshot.mjs --url http://localhost:3000<ROUTE> --output /tmp/uiux-<name>-mobile.png --width 375 --height 812

Usage — authenticated:

bash
node _uiux_screenshot.mjs --url http://localhost:3000<ROUTE> --output /tmp/uiux-<name>-desktop.png --auth /tmp/uiux-auth-state.json

After every screenshot: use the Read tool on the image file to actually see it. The image is the source of truth, not the code.

0d. Handle auth-protected screens

Check for auth middleware or providers (middleware.ts, next-auth, clerk, supabase, lucia, route guards).

If auth exists, write this to the project root and ask the user for test credentials (TEST_EMAIL, TEST_PASSWORD) and the post-login URL pattern (SUCCESS_URL, e.g. **/dashboard**):

_uiux_auth.mjs (write to project root):

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

const { TEST_EMAIL, TEST_PASSWORD, LOGIN_URL, SUCCESS_URL } = process.env;
if (!TEST_EMAIL || !TEST_PASSWORD) {
  console.error('Set TEST_EMAIL and TEST_PASSWORD'); process.exit(1);
}

const browser = await chromium.launch();
const page    = await browser.newPage();
await page.goto(LOGIN_URL ?? 'http://localhost:3000/sign-in');
await page.fill('input[type="email"]',    TEST_EMAIL);
await page.fill('input[type="password"]', TEST_PASSWORD);
await page.click('button[type="submit"]');
await page.waitForURL(SUCCESS_URL ?? '**/dashboard**', { timeout: 15000 });
await page.context().storageState({ path: '/tmp/uiux-auth-state.json' });
await browser.close();
console.log('Auth session saved → /tmp/uiux-auth-state.json');

Run from the project root:

bash
TEST_EMAIL=you@example.com TEST_PASSWORD=secret LOGIN_URL=http://localhost:3000/sign-in SUCCESS_URL="**/dashboard**" node _uiux_auth.mjs

If no test credentials are available: screenshot public screens now, mark protected screens auth-blocked, revisit them when credentials are provided.

0e. Find design tokens

Search for the token layer (packages/tokens/, tokens.css, tailwind.config.ts, CSS custom properties in a globals file). Note where spacing, colors, and font sizes are defined. Fixes that apply globally go into tokens, not components.


Phase 1 — Screen Audit

Walk the codebase and list every distinct screen. For each:

  • Route path + file path
  • One-line description
  • Auth-protected: yes/no
  • Screenshot it (with --auth flag if protected), Read the image, rate: broken / rough / decent / polished

Discover screens from:

  • App Router: every app/**/page.tsx
  • Pages Router: pages/**/*.tsx (skip _app, _document, api/)
  • Modal/drawer components
  • Empty, error, loading, not-found states
  • Auth flows (sign in, sign up, forgot password, verify email)
  • Onboarding, settings, profile, billing pages

Present the full numbered list with quality ratings before continuing.


Phase 2 — Priority Ranking

Score each screen:

FactorPoints
Entry point / first impression+3
Revenue or conversion critical+3
Highest traffic / most-seen+2
Current quality is broken or rough+2
Auth-blocked (can't screenshot yet)−1
Power-user only−1

Produce a ranked list (highest score = rank 1). Show reasoning for the top 5. Ask the user to confirm or reorder before starting Phase 3.


Phase 3 — Recursive Redesign Loop

Work through screens in ranked order. Maximum 3 iterations per screen — flag and advance if the checklist still fails after 3 passes. Track shared component fixes throughout (3f).


3a. Screenshot before

Take desktop and mobile screenshots. Read both image files. Describe in 2–3 sentences what you actually see — this is the baseline.

3b. Analyse

Read the page component, layout wrappers, child components, and CSS/Tailwind classes. Cross-reference everything against the screenshots — the screenshot overrides code assumptions. Check every dimension:

Hierarchy — One clear MIT per screen? Eye flows naturally? Color — States derived with S+B rule? 60-30-10 applied? WCAG AA met? Typography — Max 2 fonts? Heading/body ratio ≥1.75×? Weights correct? Spacing — Padding rhythm consistent? Whitespace doubled from comfortable? Shadows & depth — Light from sky? Shadow opacity matched to luminosity? Controls — Adjacent to what they affect? Unnecessary dropdowns? Text on images — Overlay / blur / floor-fade used? Alignment — Alignment lines present? Icons/bullets hanging? Empty states — Designed and instructive, not blank? Responsive — Mobile 375px unbroken in screenshot? Accessibility — Tap targets ≥44px, aria labels, color not the only signal?

3c. Issue list

Numbered, grouped by severity:

  • Critical — broken, unreadable, accessibility failure
  • Major — hierarchy wrong, color system violated, poor spacing
  • Minor — polish, animation, edge cases

3d. Fix — tokens first, then components

Before touching any component, ask: does this fix apply to ≥2 screens?

  • Yes → edit the token/config file. Do not hardcode it in the component.
  • No → edit the component directly.

Apply all fixes, critical first. Apply rules automatically — no permission needed:

  • Darker state: increase S + decrease B + shift hue toward 0°/120°/240°
  • Lighter state: decrease S + increase B + shift hue toward 60°/180°/300°
  • Neutral surfaces: brand hue whisper (~0.005 chroma), never pure gray
  • Whitespace: double until it feels like too much — that is correct
  • One MIT — only the page title gets all-out up-pop
  • Font size: body ≥15px, labels ≥12px
  • Shadow on light bg: 20–30% opacity. On dark bg: 40–50% opacity.
  • Replace dropdowns (≤5 options) with segmented, toggle, stepper, or typeahead

3e. Wait for hot reload, then screenshot

After making edits, wait for the dev server to recompile:

bash
sleep 3

Then take desktop and mobile screenshots. Read both image files. Compare visually against the before screenshots.

Answer from the images (not the code):

  1. MIT — Is the most important element the first thing the eye lands on?
  2. Flow — Does attention travel in the right order?
  3. Space — Does the screen breathe, or is it cramped?
  4. Consistency — Does this screen feel like the same product as the others?

If any answer is no: go back to 3b (increment iteration counter, max 3).

3f. Shared component tracking

When a fix touches a component used across multiple screens:

SHARED FIX: <ComponentName> — affects screens: [list routes]

When reaching those screens later: skip re-fixing, but screenshot and verify the component renders correctly in that context.

3g. Pixel-perfect checklist

Every item verified from the screenshot, not the code:

  • MIT immediately obvious when looking at the screenshot
  • Color states derived with S+B rule (no raw opacity hacks)
  • Neutral surfaces have brand hue whisper, not pure gray
  • Body text ≥15px, labels ≥12px, heading/body ratio ≥1.75×
  • Whitespace doubled — padding ≥ font-size on all sides
  • Consistent spacing rhythm (4/8/16/24/32/48px)
  • Shadow opacity matches background luminosity
  • All controls adjacent to what they affect
  • Tap targets ≥44px — verified on mobile screenshot
  • WCAG AA contrast: 4.5:1 body, 3:1 headlines
  • Empty states designed (not blank)
  • No unnecessary dropdowns
  • Icons match typeface stroke weight and corner radius
  • Text on images uses overlay / text-box / blur / floor-fade
  • Alignment lines consistent across the screen
  • Mobile 375px: no overflow, no broken layout

Any fail → go back to 3d, increment iteration. After iteration 3: flag and advance.


Phase 4 — Cleanup & Progress Report

Delete the temp scripts from the project root:

bash
rm -f _uiux_screenshot.mjs _uiux_auth.mjs

After each screen:

✓ Screen name (rank N, iteration K) — A critical + B major + C minor fixed
  Shared fixes: [list or none] | Flagged: [list or none]

After all screens, full summary table:

RankScreenRouteIssues FixedIterationsState
1............✓ polished / ⚠ review needed / 🔒 auth-blocked

List all flagged and auth-blocked screens at the bottom with specific unresolved items.


Rules

  1. Read every screenshot with the Read tool. Never describe the UI from code alone. The image wins.
  2. Tokens before components. Global fixes go in the token/config layer first.
  3. Shared components once. Fix once, track affected screens, verify in context.
  4. Max 3 iterations per screen. Flag and move on. Don't get stuck.
  5. Apply design rules automatically. No permission needed for color, spacing, or typography.
  6. Preserve functionality. Never change logic, routing, data fetching, or accessibility semantics.
  7. Mobile first. Every fix must pass at 375px before it is done.
  8. Auth-blocked screens. Screenshot what you can now. Revisit when credentials are provided.
  9. One screen at a time. Complete checklist and all iterations before advancing.
  10. Report progress. The user always knows which screen is active and how many remain.

Supported project types

Next.js (App Router or Pages), React + Vite, SvelteKit, Nuxt, Astro, plain HTML/CSS. Adjust file discovery for the project structure found.

See PRINCIPLES.md for the full design knowledge base quick reference.

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 Uiux AI skill do?

Master UI/UX redesign skill that audits every screen in the project, ranks them by priority, then redesigns each one recursively until pixel-perfect — applying color theory, typography, hierarchy, spacing, and accessibility principles automatically. Use when user wants to improve, redesign, or polish the UI/UX of an app, says "make the UI better", "redesign this", "pixel perfect", or invokes /uiux.

Why use Uiux on TypingMind?

Because you install it once and use it with any model. Uiux 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 Uiux in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/dawitlabs/ui-skills/tree/master/skills/uiux. 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 Uiux?

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 Uiux?

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

Is the Uiux 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 👇