Design AI Assets Skill
You MUST use this skill before using AI tools to generate design assets. This applies to image generation (Midjourney, DALL-E), copy assistance (Claude, ChatGPT, Gemini), and other AI-assisted design workflows.
Design AI Assets Skill
Description
You MUST use this skill before using AI tools to generate design assets. This applies to image generation (Midjourney, DALL-E), copy assistance (Claude, ChatGPT, Gemini), and other AI-assisted design workflows.
Purpose
Create production-quality assets using AI tools that:
- Match brand and design system
- Are properly licensed for use
- Meet quality standards
- Are documented for reproducibility
- Integrate seamlessly with designs
When to Use
- Generating images, illustrations, or graphics
- Creating placeholder or hero imagery
- Writing microcopy or UI text
- Generating icons or patterns
- Exploring visual directions
Prerequisites
- Clear asset requirements
- Brand guidelines and design tokens
- Understanding of AI tool capabilities
- License/usage rights verified
Process
Step 1: Define Asset Requirements
Document what you need:
## Asset Brief
### Type
Image / Illustration / Icon / Pattern / Copy
### Purpose
Hero image for landing page
### Specifications
- Dimensions: 1920x1080px
- Style: Photo-realistic / Illustrated / Abstract
- Mood: Professional, trustworthy, modern
- Color palette: Must align with brand (blue/white primary)
- Subject: Diverse team collaborating in modern office
### Usage
- Web: Landing page hero
- Social: May be cropped for social cards
- Print: None
### Restrictions
- No identifiable faces (if needed, use diverse representation)
- No competitor logos/products visible
- Must be original/licensed for commercial use
Step 2: Select Appropriate AI Tool
Choose the right tool for the job:
## AI Tool Selection
### Image Generation
| Tool | Best For | Style Strengths |
|------|----------|-----------------|
| Midjourney | Artistic, stylized, illustrations | Aesthetic quality |
| DALL-E 3 | Photorealistic, specific compositions | Prompt adherence |
| Stable Diffusion | Customizable, self-hosted | Control, privacy |
### Copy/Text
| Tool | Best For |
|------|----------|
| Claude | Nuanced copy, technical writing, tone matching |
| ChatGPT | Versatile, creative variations |
| Gemini | Multimodal, image-aware copy |
### Selected Tool
**Midjourney** - Best for our illustrated style, high aesthetic quality
Step 3: Craft Effective Prompts
Structure prompts for best results:
## Prompt Engineering
### Image Prompt Structure
[Subject] [Action/Context] [Style] [Medium] [Lighting] [Composition] [Parameters]
### Example: Team Collaboration Image
**Prompt v1:**
"Diverse team of professionals collaborating around a modern glass table,
minimalist office with plants, soft natural lighting, corporate photography style,
shot on Sony A7III, shallow depth of field --ar 16:9 --style raw --v 6"
**Prompt v2 (refined):**
"Three colleagues reviewing designs on laptop, modern coworking space,
warm natural light from large windows, professional corporate photography,
diversity in team, clean minimal aesthetic, blue and white color scheme
--ar 16:9 --style raw --s 250 --v 6"
### Parameters
| Param | Value | Effect |
|-------|-------|--------|
| --ar | 16:9 | Landscape aspect ratio |
| --style | raw | Less stylized, more realistic |
| --s | 250 | Stylization level (0-1000) |
| --v | 6 | Midjourney version |
Step 4: Iterate and Refine
Document the iteration process:
## Iteration Log
### Batch 1 (Initial)
- Prompt: [v1 prompt]
- Results: 4 variations
- Selected: None - lighting too harsh
### Batch 2 (Lighting Fix)
- Prompt: Added "soft natural lighting"
- Results: 4 variations
- Selected: #3 - good composition, needs color adjustment
### Batch 3 (Color Match)
- Prompt: Added "blue and white color scheme"
- Upscaled: Variation #2
- Selected: FINAL - matches brand, good quality
### Final Selection
- File: team-collaboration-v3-2-upscaled.png
- Resolution: 2048x1152
- Post-processing needed: Color grade to match brand primary
Step 5: Post-Process Assets
Align with design system:
## Post-Processing
### Color Correction
- Adjusted blue tones to match primary-500 (#3B82F6)
- Increased warmth by 10%
- Lifted shadows for consistency
### Cropping/Sizing
- Cropped to exactly 1920x1080
- Created social variant: 1200x630
- Created mobile variant: 750x1334
### Export
- Web: JPEG, 80% quality (optimized)
- Full: PNG, lossless (archive)
- Social: JPEG, 90% quality
Step 6: AI Copy Generation
For text and microcopy:
## Copy Generation with Claude/ChatGPT
### Brief
**Context**: Error message for failed payment
**Tone**: Friendly, helpful, not alarming
**Length**: 1-2 sentences
**Action needed**: Retry or contact support
### Prompt
"Write an error message for when a payment fails. The tone should be
friendly and helpful, not alarming. Include a suggestion to retry or
contact support. Keep it to 1-2 sentences. Provide 5 variations."
### Generated Options
1. "Your payment didn't go through this time. Please try again or contact our support team."
2. "We couldn't process your payment. Double-check your details and try again."
3. "Something went wrong with your payment. Give it another try, or we're here to help."
4. "Payment unsuccessful-this happens sometimes. Try again or reach out to us."
5. "We hit a snag processing your payment. Please retry or contact support."
### Selected: Option 3
- Friendly tone ✓
- Acknowledges issue without blame ✓
- Clear next steps ✓
Step 7: Document for Reproducibility
## Asset Documentation
### Final Asset
- File: hero-team-collaboration.jpg
- Dimensions: 1920x1080px
- File size: 245KB
### Generation Details
- Tool: Midjourney v6
- Final prompt: "[full prompt]"
- Seed: 3847291 (if available)
- Post-processing: Lightroom preset "Brand-Blue-Warm"
### License
- Generated with: Midjourney Pro subscription
- Commercial use: Allowed per Midjourney ToS
- Attribution: Not required
### Usage Rights
- Web: ✓ Unlimited
- Print: ✓ Unlimited
- Resale: ✗ Not permitted
Best Practices
Do
- Start with detailed briefs before prompting
- Iterate systematically, documenting each attempt
- Post-process to match brand colors
- Save prompts and seeds for reproducibility
- Verify licensing for commercial use
Don't
- Accept first-generation results without review
- Ignore brand guidelines in prompts
- Use AI-generated faces without consideration
- Skip post-processing for brand alignment
- Forget to document the generation process
Common Patterns
Hero Images
Style: Photorealistic or illustration matching brand
Composition: Leave space for text overlay
Color: Match brand palette
Resolution: 2x for retina displays
Illustrations
Style: Consistent across set
Characters: Diverse, inclusive representation
Background: Simple, doesn't compete
Export: SVG when possible, PNG fallback
Placeholder Content
Purpose: Design mockups, not production
Style: Match intended final content
Label: Mark clearly as placeholder
Replace: Before production launch
Output Checklist
- Asset requirements documented
- Appropriate AI tool selected
- Prompts crafted and documented
- Multiple iterations generated
- Best result selected with rationale
- Post-processing applied for brand alignment
- Final assets exported in required formats
- Full prompt/generation documented
- Licensing verified and noted
- Assets organized in design system
This skill enables effective AI-assisted asset creation with proper documentation, brand alignment, and licensing compliance.
Related Documents
AI Artist — Engineering Specification
name: ai-artist-engineering-spec
Image Generation Skill
name: image-generation
Skill: Google Stitch for Content Visuals (MCP-Integrated)
name: google-stitch-content
coinset CLI
Use this tool to query, inspect, and investigate the Chia blockchain. `coinset` wraps the Chia Full Node RPC (hosted at api.coinset.org or any full node) and includes built-in spend inspection and local CLVM utilities.