Ux Designer logo

Ux Designer

Community
szilu
ux-designer

UX/UI design guidance for building, reviewing, and critiquing interfaces and frontend code. Use when designing screens or components, auditing usability or accessibility (WCAG 2.2, EAA), writing microcopy, or designing forms, navigation, search, tables, dashboards, onboarding, notifications, real-time collaboration, canvas/whiteboard apps, AI/chat interfaces, i18n/RTL, voice, design systems, or barcode/QR scanning and display.

Overview

Publisherszilu
Repositoryux-designer-skill
Skill nameux-designer
Stars
94
Forks
15
Bundled files
29
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.

  • 29 bundled files

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

  • Open source

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

Installation

Install the Ux Designer 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/szilu/ux-designer-skill/tree/main
  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/szilu/ux-designer-skill.git \
  .claude/skills/ux-designer
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Ux Designer 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 Ux Designer 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 Ux Designer 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.

UX Designer

Use this skill for three jobs: reviewing an interface (a screenshot, a URL, or code), building UI, and advising on a design decision. You know the general UX canon already. This file gives you the workflows, the numbers, and the calls that are easy to get wrong. The reference files go deeper, so load only the ones the task needs (see the routing table).

Philosophy

  • Calm over clever. Motion, color, and density have to help the user understand something. Decoration that only impresses is noise.
  • Judgment over polish. Anyone can generate polished UI now, so the value is in correctness, research, and knowing what to leave out.
  • AI as copilot, not autopilot. AI assistance is optional, labeled, reversible, and the user stays in control.
  • Adapt to real needs. Avoid opaque or manipulative personalization.

Workflow: Reviewing or auditing a UI

  1. Establish context. Identify the users, their primary task, the platform, and any stated constraints (brand, design system, legal regime). If there's no context, assume a general audience and say so.

  2. Look at the real thing. Render it if you can: take a screenshot or open it in the browser. Walk through the primary task with the keyboard only, and run an automated checker (axe, Lighthouse) when a browser is available. Reading code alone misses contrast, overflow, focus order, and layout at other widths.

  3. Check at the extremes. Try 320px width, 200% zoom, a long translated string, empty data, huge data, a slow network, an error state, and dir="rtl" if the product is localized.

  4. Report findings ranked by user impact, not in checklist order:

    [Blocker|Major|Minor] <what is wrong> — <who it hurts and how>
      Where: <screen/component or file:line>
      Fix: <concrete change>   Ref: <heuristic / WCAG SC / reference file>

    A blocker stops a user from completing the task, or it's a legal accessibility failure. Cite WCAG success criteria by number (for example, 2.4.11 Focus Not Obscured). Skip praise padding. If a pass finds nothing serious, say that plainly.

Workflow: Building UI

  1. Reuse before inventing. Look for an existing design system, tokens, component library, and nearby screens. Match them, even where your own taste differs.
  2. Design every state, not only the happy path: empty, loading, partial, error, offline, permission-denied, and overflow (long names, 0/1/many items).
  3. Semantics first. Start from native elements (<button>, <label>, <dialog>, <details>, <input type=…>). Add ARIA only when no native element fits.
  4. Verify by looking. Render the result. Check it at mobile width, with the keyboard, and in dark mode if the product supports it. Fix what you see and check again.

Generated-UI defaults to avoid

These are telltale marks of model-generated interfaces. Don't use them unless the brand calls for them:

  • Purple/indigo gradients, glassmorphism, and glow shadows used as default styling
  • Emoji standing in for icons, and decorative icons on every heading or list item
  • A card around everything, nested cards, and every section centered with the same padding
  • Hero sections, testimonials, and marketing filler inside app UI that doesn't need them
  • Lorem-ipsum-perfect sample data: use realistic lengths, names from several locales, and edge cases
  • Buttons that all carry equal visual weight, with no single primary action
  • Gray-on-gray low-contrast "minimal" text
  • Hover-only affordances with no touch or keyboard equivalent

Core rules

Visual. Give each screen one dominant element. Body text is 16px or larger, with 1.4–1.6 line height and 50–75 characters per line. Text contrast is at least 4.5:1, or 3:1 for large text and UI component boundaries. Take all spacing from one 4px or 8px scale.

Interaction. Show visible feedback within 100ms. Most UI transitions run 150–300ms, and only large or spatial motion goes up to 500ms. Honor prefers-reduced-motion. On mobile, put primary actions in the thumb zone. Never disable a button without saying why.

Forms. Put labels above fields and never use placeholder-only labels. Validate on blur and re-validate on input once a field has shown an error. Show errors next to the field and summarize them at submit. Ask only for what you need. Mark whichever set of fields is the minority, required or optional.

Navigation. The current location is always visible. Keep the top level scannable: about 5–7 items on desktop and 3–5 in a mobile bottom bar. Miller's 7±2 is a working-memory figure, not a menu rule. On desktop, don't hide primary navigation behind a hamburger.

Accessibility. Everything is operable by keyboard, and focus is visible and never hidden behind sticky UI. Every control has an accessible name and role. Color is never the only signal. Images have meaningful alt text, or alt="" if they're decorative. Targets are at least 24×24 CSS px (WCAG 2.5.8 AA), with 44pt (iOS) or 48dp (Android) recommended on touch. Treat WCAG 2.2 AA as the legal target. WCAG 3 is still a draft.

Collaboration. Show presence (avatars, cursors, selections), make sync and offline state visible, keep undo per client, and state permission levels plainly.

Canvas. Zoom around the cursor, not the screen center. Snapping has a toggle. Provide a minimap for large boards, full keyboard navigation, and viewport culling.

AI. Label AI content, attribute sources, and provide stop/cancel, edit/regenerate, feedback, and undo for any AI-applied change. Keep a human override.

Onboarding. Guide the user to the first real value, not a tour. Make it skippable and never show it again once finished. Empty states give the next action.

Notifications. Match visual severity to real severity. Ask for push permission in context, after the user has seen value. Give users per-channel control. Toasts auto-dismiss after 4–8s, except toasts that carry an action, which stay until dismissed (WCAG 2.2.1).

Ethics. Accept and reject get equal prominence. Optional consent boxes start unchecked. Cancelling is as easy as signing up. No confirmshaming.

i18n. Use logical CSS properties and verify the layout in dir="rtl". Allow 30–40% text expansion and no fixed-width labels. Externalize all strings and keep text out of images. Format dates, numbers, and currency with Intl, and handle plurals with ICU rules. Language switchers use endonyms, not flags.

Decision trees

Modal vs. side panel vs. full page

Quick confirmation or 1-3 fields?                 → Modal
Edit details while keeping context visible?
  narrow content (form, properties, chat)         → Side panel
  needs width                                     → Full-page overlay with back
Multi-step: short steps → modal + stepper; long steps / needs reference → full page + stepper
Creating a complex entity (document, project)?    → Full page

Notification type

Blocking, must resolve now                        → Modal
Urgent, non-blocking                              → Banner, persistent until dismissed
Completed action: success/info                    → Toast, auto-dismiss 4-8s
Completed action: warning/error or has action     → Toast, manual dismiss
Background event, same context                    → Badge + inline indicator
Background event, elsewhere                       → Nav badge (+ optional push)
System status (maintenance, connectivity)         → Persistent banner

Reference routing

Load the file when the task touches its topic. Each file opens with a table of contents, so you can jump to the section you need.

Load when the task involves…File
Heuristic evaluation, Nielsen's 10, Gestaltreferences/01-core-principles.md
Fitts, Hick, Jakob, Tesler, peak-end, and similar lawsreferences/02-laws-of-ux.md
WCAG 2.2 criteria, ARIA, EAA / legal compliancereferences/03-accessibility.md
Typography, color, spacing, hierarchy, dark modereferences/04-visual-design.md
Navigation structure, sitemaps, card sortingreferences/05-information-architecture.md
Modals, tooltips, drag-and-drop, motion timingreferences/06-interaction-design.md
Form layout, validation, input types, errorsreferences/07-forms-and-inputs.md
Touch targets, gestures, responsive, mobile navreferences/08-mobile-ux.md
Microcopy, error messages, voice and tonereferences/09-ux-writing.md
Interviews, usability tests, surveys, metricsreferences/10-user-research.md
Tokens, component APIs, design system docsreferences/11-design-systems.md
Live cursors, avatars, typing/presence indicatorsreferences/12a-presence-awareness.md
Conflicts, sync, offline, sharing, version historyreferences/12b-conflict-resolution-sync.md
Canvas zoom, pan, selection, manipulationreferences/13a-canvas-navigation.md
Canvas layers, snapping, LOD, rendering performancereferences/13b-canvas-objects-performance.md
Chat UI, copilots, agents, generative UIreferences/14-ai-ux-patterns.md
Dark patterns, consent, DSA/GDPR UI rulesreferences/15-ethical-design.md
First-run, activation, empty states, checklistsreferences/16-onboarding.md
Notification systems, push, toasts, preferencesreferences/17-notifications.md
Charts, dashboards, accessible data vizreferences/18-data-visualization.md
Search, autocomplete, filters, zero resultsreferences/19-search-ux.md
Delight, trust, tone, error recovery emotionreferences/20-emotional-design.md
Tables, sorting, pagination, bulk actionsreferences/21-data-tables.md
Loading, skeletons, optimistic updates, Core Web Vitalsreferences/22-performance-ux.md
Localization, RTL, Intl, plurals, text expansionreferences/23-internationalization.md
Voice, multimodal, cross-device inputreferences/24-voice-and-multimodal.md
Camera barcode/QR scanning on phone or web, viewfinder, QR safetyreferences/25a-barcode-camera-scanning.md
Scan-driven apps: warehouse, POS, healthcare, rugged or wedge scanners, GS1references/25b-barcode-scanner-workflows.md
Showing barcodes/QR: tickets, wallet passes, QR login, printed labelsreferences/25c-barcode-display.md

Reference values

MetricValueSource / note
Target size≥ 24×24 CSS px (AA); 44pt iOS / 48dp AndroidWCAG 2.5.8; Apple HIG; Material
Body text≥ 16px, line height 1.4–1.6WCAG 1.4.12 tests up to 1.5
Line length50–75 characters
Text contrast4.5:1 normal, 3:1 large (≥ 24px, or ≥ 18.66px bold)WCAG 1.4.3
Non-text contrast3:1 (UI boundaries, focus rings, icons)WCAG 1.4.11
Feedback latency< 100ms feels instant; > 1s show a spinner; > 10s show progress with cancelNielsen response-time limits
Transitions150–300ms typical, ≤ 500ms for large motionMaterial motion
Loading indicator delay~300ms before showing, to avoid flashes
Toast4–8s auto-dismiss (no actions)
Text expansion30–40% (DE/FI/RU), up to 200%+ for short stringsW3C i18n
Working memory~4±1 chunks (Cowan); 7±2 (Miller) is dated
Canvascursor updates 50–100ms, 60fps pan/zoom, 2–8px snap threshold, 10%–4000% zoomFigma-class tools
Avatar stack3–5 visible, then "+N"
AI responsefirst token < 1s, or show immediate progress
Displayed QR code4-module quiet zone; module ≥ 0.25 mm physical; error correction M by defaultISO/IEC 18004; IATA BCBP; DENSO WAVE
QR size vs. distance≥ 2×2 cm, +1 cm per 10 cm of scan distanceNN/g heuristic

Numbers like conversion rates, NPS targets, and completion percentages depend on context. Don't quote benchmarks as universal facts. Recommend measuring against the product's own baseline.

Anti-patterns

Each one points to the reference file that covers the fix.

  • Dark patterns, confirmshaming, asymmetric consent → 15
  • Hidden desktop navigation, no sense of location → 05
  • Infinite scroll without position or a footer → 21
  • Autoplaying media, color-only signals, invisible focus → 03
  • Disabled buttons with no explanation, modal overuse → 06
  • Walls of text, no hierarchy → 04
  • Tiny touch targets → 08
  • Missing loading, empty, or error states → 22
  • Silent sync failures, no offline indication → 12b
  • Cursor overload, no presence → 12a
  • Screen-center zoom → 13a
  • Hidden AI, AI changes applied without consent or undo → 14
  • Mandatory long tours → 16
  • Notification carpet bombing, push permission on first visit → 17
  • Hardcoded strings, fixed widths, LTR-only layout → 23
  • Voice-only flows, hidden mic, no recognition feedback → 24
  • Scan-only flows with no manual entry, silent scan failures → 25a / 25b

Sources

Laws of UX · Nielsen Norman Group · WCAG 2.2 · Material Design · Apple HIG · Baymard Institute · Google PAIR · Microsoft HAX · Deceptive Design · EU DSA · EU Accessibility Act · W3C i18n · Liveblocks · Figma Engineering · Tufte · ColorBrewer · web.dev

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 Ux Designer AI skill do?

UX/UI design guidance for building, reviewing, and critiquing interfaces and frontend code. Use when designing screens or components, auditing usability or accessibility (WCAG 2.2, EAA), writing microcopy, or designing forms, navigation, search, tables, dashboards, onboarding, notifications, real-time collaboration, canvas/whiteboard apps, AI/chat interfaces, i18n/RTL, voice, design systems, or barcode/QR scanning and display.

Why use Ux Designer on TypingMind?

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

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/szilu/ux-designer-skill/tree/main. 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 Ux Designer?

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 Ux Designer?

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

Is the Ux Designer AI skill free?

Yes. It is published on GitHub by szilu 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 👇