Douyin Video Content Structured Extraction

Problem

Information in Douyin videos is often scattered across titles, hashtags, subtitles, chapter highlights, comments, and engagement metrics. Copying a link alone does not produce an archivable text record. This skill targets engineering and knowledge-management workflows, converting short-video content into structured Markdown notes without downloading, storing, or reposting the video file.

How It Works

• Link validation: recognizes v.douyin.com short links, douyin.com/video normal links, and douyin.com/live replay links; failures report specific reasons such as invalid links, private videos, or regional restrictions.
• Page reading: uses a browser session to read snapshots, then extracts fields with @ref, snapshot -i, get text, and batch --bail; references must be refreshed after DOM changes.
• Content extraction: prioritizes metadata, chapter highlights, and embedded subtitles; if subtitles are missing, it marks speech transcription as unavailable; comments or search may be used optionally when permissions allow.
• Structured output: supports a standard full version, minimal summary, quote extraction, and data summary format; the default output emphasizes detailed breakdowns, source versus extension labeling, scenario comparisons, and self-check lists.

Boundaries

This skill fits content organization for knowledge videos, parenting education, workplace tutorials, product reviews, news, and emotional storytelling. It should stop and explain when pages require login or when videos are private, paid, or permission-restricted. It does not download, store, or forward original video files, and it should not invent content that is not present in the video.