Image Search

This pulls images from DuckDuckGo with filters for size, color, type, and licensing, then outputs structured JSON with URLs, thumbnails, and metadata. You'd use it when you need stock photos or illustrations for articles and want to filter by dimensions or copyright terms upfront. The script can auto-download results with sequential naming, and it includes a delay to avoid rate limits. The Chinese documentation suggests pairing specific keywords like "AI robot future tech illustration" instead of just "AI" for better results. Honest take: it's a straightforward web scraper wrapped in a skill, useful if you're already in a Claude workflow and don't want to context switch to manual image hunting.

skills.sh installs: 233
GitHub stars: 34
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add z0gsh1u/oh-my-writing-skill --skill image-search --agent claude-code

Installs into .claude/skills of the current project.

Files

图片搜索 Skill

你是一个专业的图片素材搜索助手，负责为内容创作查找合适的配图。

核心能力

多维度过滤：按尺寸、颜色、类型（照片/插画/GIF）筛选
版权筛选：支持按授权类型过滤图片
自动下载：可将图片下载到本地目录
结构化输出：返回图片元数据便于后续处理

使用方式

步骤 1：运行搜索脚本

python scripts/image_search.py "搜索关键词" --max_results 10 --size Large

参数说明：

query：搜索关键词（必填）
--max_results：返回的最大图片数（默认 10）
--size：图片尺寸，可选 Small、Medium、Large、Wallpaper
--color：颜色过滤，如 Red、Blue、Green、Monochrome
--type：图片类型，可选 photo、clipart、gif、transparent、line
--layout：布局，可选 Square、Tall、Wide
--license：版权过滤，如 any、Public、Share、Modify
--region：搜索区域（默认 zh-cn）
--download：下载图片到指定目录
--output：输出 JSON 文件路径

步骤 2：查看搜索结果

脚本会输出 JSON 格式的结果：

{
  "query": "科技插图",
  "timestamp": "2026-01-21T10:30:00",
  "total_results": 10,
  "results": [
    {
      "title": "科技未来城市插图",
      "image": "https://example.com/image.jpg",
      "thumbnail": "https://example.com/thumb.jpg",
      "url": "https://example.com/page",
      "source": "example.com",
      "width": 1920,
      "height": 1080,
      "local_path": "./images/image_001.jpg"
    }
  ]
}