| name | image-search |
| description | Search and download UI screenshots from the web using image search. Use when you need reference screenshots of real websites/apps (e.g., Gmail, Slack, Notion) for building mock applications. Downloads high-resolution images to a local directory. |
Image Search — Download UI Reference Screenshots
Search Google/Bing Images and download results to a local directory. Ideal for collecting UI reference screenshots of real applications when building mocks.
Usage
python3 .claude/skills/image-search/scripts/image_search.py "<query>" <output_dir> [options]
Examples
python3 .claude/skills/image-search/scripts/image_search.py "gmail inbox UI screenshot" gmail_mock/assets/screenshots --max 8
python3 .claude/skills/image-search/scripts/image_search.py "slack channels sidebar desktop" slack_mock/assets/screenshots --max 5
python3 .claude/skills/image-search/scripts/image_search.py "slack thread panel reply" slack_mock/assets/screenshots --max 5
python3 .claude/skills/image-search/scripts/image_search.py "notion page editor" notion_mock/assets/screenshots --max 5 --json
Options
| Option | Default | Description |
|---|
--max N | 8 | Maximum number of images to download |
--engine | bing | Search engine: bing (recommended) or google |
--min-width | 400 | Minimum image width in pixels |
--min-height | 300 | Minimum image height in pixels |
--json | off | Output results as JSON |
Recommended search queries for mock app research
For best results, use specific queries combining the app name with UI-specific terms:
"<app> inbox UI screenshot" — main view"<app> settings page interface" — settings"<app> compose new message dialog" — modals/dialogs"<app> sidebar navigation desktop" — navigation"<app> dark mode interface" — dark theme variant"<app> mobile vs desktop UI comparison" — responsive design
Notes
- Bing is the default engine (Google blocks automated crawling)
- Downloads large, high-resolution photos by default
- If Bing fails, the script automatically falls back to Google
- Images are saved as numbered JPGs:
000001.jpg, 000002.jpg, etc.
- Requires
icrawler Python package (pip3 install icrawler)
