image-search
GitHub通过Google图片搜索查找图像、缩略图及原始来源页面,提取元数据并验证许可。遇到验证码时立即移交用户处理。
Trigger Scenarios
Install
npx skills add yetanotherbrowser/yab --skill image-search -g -y
SKILL.md
Frontmatter
{
"name": "image-search",
"sites": [],
"keywords": [
"image search",
"google images",
"find images",
"photo search",
"thumbnail",
"поиск картинок",
"найди картинки",
"найди изображения"
],
"description": "Find images, thumbnails and original source pages using Google Images in Yab."
}
Image search
Search in a task tab using https://www.google.com/search?udm=2&q= plus the encoded query.
Use the user's language and requested subject; optional safe=active enables SafeSearch.
Read google-search for navigation, consent and unusual-traffic handling.
Read result titles and source links with snapshot/text. Screenshot when the visual content matters.
Open promising results to inspect the image and its source page. Record only metadata actually
exposed there: title, source page URL, original image URL, thumbnail URL and width/height.
Use eval to read visible image/link attributes when refs do not expose those values; do not
invent original URLs or infer original dimensions from a displayed thumbnail. Some results load
only after scrolling or selecting the thumbnail.
Show a small relevant selection with links to the source pages. Label a thumbnail as a thumbnail; prefer the original URL when the user needs an asset. Verify any requested reuse license on the source page: a Google filter or a licensable label is only a hint. Download only files needed by the user's task, using Yab's download flow, and preserve attribution where supplied.
Yab has no imageSearch global. Use the existing browser tools; a captcha is the user's:
handoff at once, never try to solve it.
Version History
- a4661df Current 2026-11-16 19:55


