Douyin Scraper

Search Douyin videos by natural language — "搜索一下海鲜视频" just works.

What This Skill Does

Search and retrieve Douyin (抖音) video content using natural language queries:

• Keyword search: Find videos by any Chinese or English keyword phrase
• Trending discovery: Find what's hot on Douyin
• Video detail: Extract metadata from a specific Douyin video URL

How It Works

This skill uses two layers:

1. Web search (Brave) → discovers Douyin search result pages and video URLs
2. Web fetch → extracts readable content from Douyin pages when accessible

No API keys, cookies, or login required.

Activation Rules (for AI agents)

Use this skill when the user asks about:

• Searching Douyin videos by keyword (搜索抖音/搜一下/找一下)
• Finding trending Douyin content (抖音热搜/抖音热门)
• Getting info about a specific Douyin video URL
• Any request containing "抖音" + a search intent

Do NOT use this skill when:

• User wants to download videos (use a downloader skill instead)
• User wants to publish/upload to Douyin
• User asks about Douyin script optimization (that's the douyin skill)
• User asks about Douyin account/analytics management

Execution Protocol

Search: douyin_search

Trigger: User wants to search Douyin content by keyword.

Steps:

1. Extract keywords from the user's natural language request.

   • "搜索一下海鲜视频" → keyword: 海鲜视频
   • "帮我找抖音上做菜的视频" → keyword: 做菜
   • "抖音上最近火的猫咪内容" → keyword: 猫咪
   • "douyin seafood cooking" → keyword: seafood cooking

2. Run web_search with the keyword targeting Douyin:

   web_search(query="\x3Ckeyword> site:douyin.com", count=10)
   

   Also run a broader search for richer results:

   web_search(query="抖音 \x3Ckeyword> 视频", count=10)
   

3. Parse results — extract from search results:

   • Video URLs: https://www.douyin.com/video/\x3Cid>
   • Search pages: https://www.douyin.com/search/\x3Ckeyword>
   • Descriptions and snippets from search result titles/descriptions

4. Enrich — for each distinct video URL found, optionally use web_fetch to extract more detail. If the page is behind CAPTCHA, fall back to search result metadata only.

5. Format output as a clean list:

   🔍 抖音搜索：海鲜视频
   ━━━━━━━━━━━━━━━━━━━━━
   
   1. **盘点捞汁小海鲜** — 看哪位主播吃的最过瘾
      🔗 https://www.douyin.com/search/捞汁小海鲜
   
   2. **全网最火主播吃海鲜合集**
      🔗 https://www.douyin.com/search/全网最火主播吃海鲜合集
   
   ...
   

Trending: douyin_trending

Trigger: User asks about Douyin trending/hot topics (抖音热搜/热门/趋势).

Steps:

1. Run web_search:

   web_search(query="抖音热搜榜 today", count=10, freshness="day")
   

2. Parse and format the trending topics.

Video Detail: douyin_detail

Trigger: User provides a specific Douyin video URL or wants info about one.

Steps:

1. Try web_fetch(url=\x3Cdouyin_video_url>) to extract metadata.
2. If blocked by CAPTCHA, try web_search(query="site:douyin.com \x3Cvideo_id>").
3. Return whatever metadata is available (title, author, description).

Output Format

Always include:

• Search keyword used
• Result count found
• Each result with: title, brief description (if available), link

For searches with no results:

• Say so clearly
• Suggest trying a different or shorter keyword
• Suggest browsing directly at https://www.douyin.com/search/\x3Ckeyword>

Limitations

• Douyin aggressively blocks automated access (CAPTCHA, auth requirements)
• Video metadata may be limited to what search engines have indexed
• Cannot download videos — only discover and describe them
• Search results depend on search engine coverage, which may lag behind live Douyin
• For real-time or comprehensive search, the user should browse Douyin directly

Tips

• Shorter keywords work better — 海鲜 over 海鲜视频大全最新
• Chinese keywords yield more results than English for Douyin content
• If a search returns few results, try the Chinese equivalent of the term
• Combine with the douyin script optimization skill for content creation workflows