AnyCrawl YouTube Video Extractor

Use this skill when the user gives a public YouTube watch URL and wants deep video metadata through AnyCrawl.

Workflow

1. Accept a single public YouTube watch URL.
2. Validate that it is one of the supported shapes:
   • https://www.youtube.com/watch?v=VIDEO_ID
   • https://youtu.be/VIDEO_ID
3. Run the bundled extractor:

node .claude/skills/anycrawl-youtube-video-extractor/scripts/run.js "<youtube-watch-url>"

4. Read the JSON file path printed by the script.
5. Summarize the important fields for the user:
   • title
   • channel metadata
   • publish and upload dates
   • duration
   • views and likes
   • description
   • transcript availability
   • endscreen recommendations
   • age restriction flag
   • screenshot URL

Output Handling

The script writes a full JSON artifact to /tmp by default and prints a compact summary to stdout.

The artifact includes:

• normalized: structured YouTube-focused extraction
• anycrawl: raw AnyCrawl response, including markdown, html, links, and screenshot@fullPage when available
• request: the exact payload used for the successful attempt

If you need the official template notes, read references/template-notes.md.