imagesearchSafety 95Repository ShareFavorite skill

Instructions

Search the web for visual information, download and analyze relevant images, and synthesize findings.

Usage

/imagesearch <query>
/imagesearch majormud color system
/imagesearch "retro BBS door game UI design"
/imagesearch what font does discord use

Workflow

1. WebSearch for the query + relevant visual terms (add "screenshot", "UI", "design", "example" etc. as appropriate)

2. WebFetch promising results to extract:

   • Text information about the topic
   • Image URLs from markdown ![alt](url) tags

3. Download images to /tmp/imagesearch-$$ using curl:

   curl -sL "https://example.com/image.png" -o /tmp/imagesearch-$$/image-001.png
   

4. Read the downloaded images to analyze them visually

5. Synthesize findings into actionable information:

   • What you learned from text sources
   • What you observed in the images
   • Implementation recommendations if requested

Tips

• Add visual search terms to queries: "screenshot", "example", "UI", "design", "color palette"
• Prioritize images from documentation, wikis, and official sources
• Skip tiny images (icons, avatars) - look for screenshots and diagrams
• If a site blocks fetching, try alternative sources
• Clean up temp files when done: rm -rf /tmp/imagesearch-$$/

Image Analysis Focus

When analyzing images, pay attention to:

• Color palettes and hex values
• Typography and font choices
• Layout and spacing patterns
• UI component styles
• Visual hierarchy
• Any text visible in screenshots

Example Session

User: /imagesearch majormud color system

Claude:
1. Searches "majormud color system screenshot ANSI"
2. Fetches wiki pages, finds image URLs
3. Downloads screenshots to /tmp
4. Analyzes the visual style:
   - ANSI 16-color palette
   - Cyan for system messages
   - Yellow for item names
   - Red for combat/damage
   - etc.
5. Provides implementation recommendations

Limitations

• Some sites block WebFetch (403 errors) - try alternative sources
• Very large images may take time to download
• Can only analyze static images, not videos/animations
• Google Images directly is not accessible (anti-bot) - we find images through regular page content