Agent skill
ui-implement
Patterns for implementing UI improvements based on design analysis. Works with review documents from dev:ui-analyse or /dev:ui command. Includes Anti-AI design rules and visual verification.
Install this agent skill to your Project
npx add-skill https://github.com/MadAppGang/claude-code/tree/main/plugins/dev/skills/design/ui-implement
SKILL.md
UI Implementation Skill
Overview
This skill provides patterns for implementing UI improvements based on design analysis. It transforms review findings into code changes following Anti-AI design principles.
Relationship to Other Skills
| Skill | Purpose | Modifies Code? |
|---|---|---|
| dev:ui-analyse | Visual analysis, issue detection | No |
| dev:ui-implement | Apply improvements from analysis | Yes |
| dev:ui-style-format | Style file specification | No |
| dev:design-references | Reference image management | No |
Prerequisite
This skill assumes analysis has been completed using:
dev:ui-analyseskill/dev:uicommand output- External design review
Before implementing, ensure you have:
- A review document with identified issues
- Component path(s) to modify
- Understanding of the visual metaphor to apply
Implementation Workflow
Phase 1: Parse Review Document
Extract actionable items from the review document.
# Extract actionable items from review
parse_review() {
local review_path="$1"
# Extract CRITICAL and HIGH issues
grep -A 2 "^\[CRITICAL\]" "$review_path"
grep -A 2 "^\[HIGH\]" "$review_path"
}
What to Look For:
- CRITICAL issues (fix immediately)
- HIGH issues (fix before release)
- Specific component locations
- Recommended changes
Phase 2: Visual Context (Optional)
If Gemini available and screenshot provided, understand the current visual state.
# Understand current visual state
npx claudish --model "$GEMINI_MODEL" --image "$CURRENT_SCREENSHOT" --quiet --auto-approve <<< "
Describe the current UI implementation:
1. Layout structure
2. Color scheme
3. Typography
4. Animation presence
5. Areas needing improvement
Output as structured data for implementation."
Provider Detection:
# Check providers in priority order
if [[ -n "$GEMINI_API_KEY" ]]; then
GEMINI_MODEL="g/gemini-3-pro-preview"
elif [[ -n "$OPENROUTER_API_KEY" ]]; then
GEMINI_MODEL="or/google/gemini-3-pro-preview"
elif [[ -n "$GOOGLE_APPLICATION_CREDENTIALS" ]]; then
GEMINI_MODEL="vertex/gemini-3-pro-preview"
else
GEMINI_MODEL="" # Text-only mode
fi
Phase 3: Apply Anti-AI Improvements
Apply the five core Anti-AI design rules to transform generic UI into distinctive design.
See Anti-AI Design Rules section below for detailed patterns.
Phase 4: Visual Verification (Optional)
If Gemini available after changes, verify improvements were applied correctly.
# Verify improvements
npx claudish --model "$GEMINI_MODEL" --image "$NEW_SCREENSHOT" --quiet --auto-approve <<< "
Verify these improvements were applied:
1. Asymmetric layout: {expected}
2. Texture/depth: {expected}
3. Typography drama: {expected}
4. Micro-interactions: {expected}
5. Bespoke colors: {expected}
Score 1-10 and note any remaining issues."
Anti-AI Design Rules
These rules transform generic, AI-generated looking UI into distinctive, human-crafted designs.
Rule 1: Break Symmetry
Rigid symmetric grids look AI-generated. Break them with asymmetric bento layouts.
// BEFORE: Symmetric grid
<div className="grid grid-cols-3 gap-4">
// AFTER: Asymmetric bento
<div className="grid grid-cols-12 gap-6">
<div className="col-span-7 row-span-2" />
<div className="col-span-5" />
<div className="col-span-3" />
<div className="col-span-2 -mt-8" />
</div>
Key Techniques:
- Use 12-column grids for flexibility
- Vary column spans (7+5, not 6+6)
- Use negative margins for overlap
- Vary row heights
- Break alignment intentionally
Rule 2: Add Texture
Flat solid colors look AI-generated. Add gradients, glass effects, and depth.
// BEFORE: Flat
<div className="bg-white rounded-lg">
// AFTER: Textured
<div className="
bg-gradient-to-br from-white/80 to-white/40
backdrop-blur-xl
border border-white/20
shadow-[0_8px_32px_rgba(0,0,0,0.08),inset_0_1px_0_rgba(255,255,255,0.6)]
">
Key Techniques:
- Gradient backgrounds (subtle, directional)
- Glassmorphism (backdrop-blur + semi-transparent)
- Complex shadows with inset highlights
- Border with alpha for subtle edges
- Noise texture overlays
Rule 3: Dramatic Typography
Generic typography looks AI-generated. Create dramatic visual hierarchy.
// BEFORE: Generic
<h1 className="text-2xl font-bold">Welcome</h1>
// AFTER: Dramatic
<h1 className="
text-[clamp(4rem,15vw,12rem)]
font-serif font-thin tracking-[-0.04em]
leading-[0.85]
bg-gradient-to-r from-zinc-900 via-zinc-600 to-zinc-900
bg-clip-text text-transparent
">
Welcome
</h1>
Key Techniques:
- Fluid typography with clamp()
- Extreme size contrasts (body 16px, hero 120px+)
- Tight line-height for headlines (0.85-0.95)
- Negative letter-spacing for large text
- Gradient text with bg-clip-text
- Mix serif/sans-serif for contrast
Rule 4: Micro-Interactions
Static elements look AI-generated. Add motion and feedback.
// BEFORE: Static
<button className="bg-blue-500 hover:bg-blue-600">
// AFTER: Reactive
<motion.button
whileHover={{ scale: 1.02, y: -2 }}
whileTap={{ scale: 0.98 }}
transition={{ type: "spring", stiffness: 400, damping: 17 }}
className="
bg-gradient-to-r from-violet-600 to-indigo-600
hover:shadow-[0_0_40px_rgba(139,92,246,0.4)]
transition-shadow duration-300
"
>
Key Techniques:
- Spring physics for natural feel
- Subtle scale on hover (1.02-1.05)
- Y-axis lift effect
- Shadow glow expansion
- Staggered animations for lists
- Exit animations (not just enter)
Rule 5: Bespoke Colors
Default Tailwind colors look AI-generated. Create custom palettes.
// BEFORE: Default palette
<div className="bg-blue-500 text-white">
// AFTER: Bespoke palette
<div className="
bg-[#0D0D0D]
text-[#E8E4DD]
bg-gradient-to-br
from-[#1a1a2e] via-[#16213e] to-[#0f3460]
">
Key Techniques:
- Custom hex colors (not Tailwind defaults)
- Off-whites instead of pure white (#E8E4DD, #FAF9F6)
- Rich blacks instead of pure black (#0D0D0D, #1a1a2e)
- Color gradients across multiple stops
- Consistent palette derived from style guide
Visual Metaphor Library
When implementing, select an appropriate visual metaphor to guide decisions.
| Metaphor | Use Case | Key Characteristics |
|---|---|---|
| Cyberpunk Glass | Dashboards, tech | Neon accents + glassmorphism + dark backgrounds |
| Swiss Minimalist | Professional, B2B | Strict grid + high contrast + precise spacing |
| Neo-Brutalism | Creative, bold | Thick borders + clashing colors + raw shapes |
| Organic Luxury | Premium, fashion | Warm neutrals + serif typography + slow motion |
| Editorial Magazine | Marketing, content | Large display type + asymmetric images + whitespace |
Applying a Metaphor:
- Choose based on brand/context
- Apply consistent characteristics across all changes
- Document choice in implementation log
- Verify consistency in visual check
Required Dependencies
Ensure these are installed before implementing micro-interactions:
npm install framer-motion lucide-react
# or
bun add framer-motion lucide-react
Dependency Usage:
framer-motion: All animation and gesture handlinglucide-react: Icon library (consistent, customizable)
Implementation Log Format
When completing implementation, create a log documenting all changes.
## Implementation Log
**Component**: {component_path}
**Session**: {session_id}
**Date**: {timestamp}
### Changes Applied
1. **{Issue from Review}**
- Before: {description}
- After: {description}
- Code: {file:line}
2. **{Issue from Review}**
- Before: {description}
- After: {description}
- Code: {file:line}
### Visual Metaphor
{metaphor_name}: {brief explanation of why chosen and how applied}
### Anti-AI Rules Applied
- [x] Rule 1: Break Symmetry - {how applied}
- [x] Rule 2: Add Texture - {how applied}
- [x] Rule 3: Dramatic Typography - {how applied}
- [ ] Rule 4: Micro-Interactions - {not applicable or how applied}
- [x] Rule 5: Bespoke Colors - {how applied}
### Dependencies Added
{list if any, or "None"}
### Verification Status
- [ ] Visual verified with Gemini
- [ ] Manual testing recommended
- [ ] Responsive breakpoints checked
- [ ] Accessibility preserved
Best Practices
DO
- Read the full review document before starting
- Apply changes systematically (one rule at a time)
- Preserve existing accessibility features
- Test responsive behavior after changes
- Document all changes in implementation log
- Use Gemini verification when available
DON'T
- Apply all Anti-AI rules blindly (assess what's needed)
- Break existing functionality for aesthetics
- Ignore accessibility for visual appeal
- Skip the implementation log
- Forget to install required dependencies
- Mix multiple visual metaphors inconsistently
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
test-skill
A test skill for validation testing. Use when testing skill parsing and validation logic.
bad-skill
claudish-usage
CRITICAL - Guide for using Claudish CLI ONLY through sub-agents to run Claude Code with OpenRouter models (Grok, GPT-5, Gemini, MiniMax). NEVER run Claudish directly in main context unless user explicitly requests it. Use when user mentions external AI models, Claudish, OpenRouter, or alternative models. Includes mandatory sub-agent delegation patterns, agent selection guide, file-based instructions, and strict rules to prevent context window pollution.
release
Plugin release process for MAG Claude Plugins marketplace. Covers version bumping, marketplace.json updates, git tagging, and common mistakes. Use when releasing new plugin versions or troubleshooting update issues.
claudish-integration
openrouter-trending-models
Fetch trending programming models from OpenRouter rankings. Use when selecting models for multi-model review, updating model recommendations, or researching current AI coding trends. Provides model IDs, context windows, pricing, and usage statistics from the most recent week.
Didn't find tool you were looking for?