Agent skill
web-ui-ux
Web UI/UX specialist guidance for designing, reviewing, and polishing web product UI (layout, usability, microcopy, accessibility, responsive behavior, forms, navigation). Use when asked to improve UI/UX, audit a page, design a screen, create a component spec, or generate HTML head/manifest/icon guidance for a web app; optionally applicable to Unreal UI (UMG) for UX heuristics.
Install this agent skill to your Project
npx add-skill https://github.com/ssdeanx/AI-Skills/tree/main/skills/web-ui-ux
SKILL.md
Web UI/UX
You help produce modern, usable, accessible web UI with clear, testable guidance.
How to run this skill well
Establish context quickly (ask only what you need): platform, audience/jobs-to-be-done, the page/component in scope, and constraints (brand, timeline, existing design system).
Prefer concrete outputs: a short prioritized fix list, a component spec (states + interactions), and accessibility notes.
Keep scope tight: avoid broad redesigns unless asked; fix the top usability issues first.
Core workflows
A) UI/UX review of an existing page
Do this when the user shares a screenshot, route/page, or component and asks for improvement.
Steps:
- Identify intent and primary action.
- Check clarity: hierarchy, CTA prominence, labels, and visual noise.
- Check usability: forms, errors, loading/empty states, and navigation.
- Check accessibility basics: keyboard, focus order, labels, contrast.
- Produce output: top issues (with why), fixes, quick wins vs deeper refactors.
Use detailed checklists in:
references/ui-review-checklist.mdreferences/accessibility-checklist.md
B) Designing a new screen/component
Do this when the user asks to design a new page, dialog, component, or flow.
Steps:
- Ask for minimum inputs: users + goal, must-have fields/actions, and target content density.
- Propose an IA/layout: structure, regions, hierarchy.
- Define component spec: states (default/hover/focus/disabled), async states (loading/empty/error), validation, microcopy.
- Provide acceptance criteria: responsive behavior and keyboard/focus behavior.
Use system defaults and token guidance in:
references/design-system-defaults.md
Output templates
UI/UX findings template
Use this structure:
- Summary (1-2 sentences)
- Top issues (prioritized)
- Issue: ...
- Why it matters: ...
- Fix: ...
- Accessibility notes
- Responsive notes
- Copy/microcopy suggestions
Component spec template
Use this structure:
- Purpose
- Anatomy (slots/parts)
- States (default/hover/focus/disabled/loading/error/empty)
- Keyboard interactions
- Validation rules (if form-related)
- Responsive behavior
Notes
- Keep content ASCII-friendly when possible to avoid Windows encoding pitfalls in older validators.
- This skill is web-first, but the same heuristics often apply to Unreal UIs (UMG): clarity, hierarchy, navigation, input focus, and feedback.
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
task-generator
Generate structured task lists from specs or requirements. IMPORTANT: After completing ANY spec via ExitSpecMode, ALWAYS ask the user: "Would you like me to generate a task list for this spec?" Use when user confirms or explicitly requests task generation from a plan/spec/PRD.
icon-generator
Generate web UI/UX icon assets (favicon.ico, apple-touch-icon, PWA icons incl. maskable) and optionally Unreal Engine packaging icons (Windows .ico, macOS .iconset/.icns, Linux .png) from a single source SVG/PNG; use when you need correct multi-size icon files, safe-area guidance, manifests/head tags, or automation.
hello-world
A simple introductory skill that demonstrates basic OpenCode skill functionality and provides a starting template
skill-creator
Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.
verl-rl-training
Provides guidance for training LLMs with reinforcement learning using verl (Volcano Engine RL). Use when implementing RLHF, GRPO, PPO, or other RL algorithms for LLM post-training at scale with flexible infrastructure backends.
openrlhf-training
High-performance RLHF framework with Ray+vLLM acceleration. Use for PPO, GRPO, RLOO, DPO training of large models (7B-70B+). Built on Ray, vLLM, ZeRO-3. 2× faster than DeepSpeedChat with distributed architecture and GPU resource sharing.
Didn't find tool you were looking for?