Clip Media Extractor

What it solves

When a user provides only a link from YouTube, TikTok, Bilibili, Instagram, or another supported page, the agent still needs the title, metadata, downloadable file path, or a temporary share link. clip-media turns that request into one script call, reducing the need to manually assemble download, size routing, upload, and reply steps.

How it works and where it stops

The skill relies on yt-dlp, ffmpeg, and curl. It generally analyzes the URL to detect the platform and media type, downloads the highest-quality version, then routes the result by size. Files under 50 MB return a local path that the agent can send directly; files over 50 MB default to temporary public hosts such as tmpfiles or transfer and return a download link. Use --title for title or description only, --info for JSON metadata, --test to validate a URL, --list-sites to inspect supported platforms, and --no-upload to skip cloud upload completely. Keep in mind: files over 50 MB are uploaded to public third-party storage where anyone with the link can access them, so the skill should not be used for private, paid, regulated, or copyright-sensitive content. --cookies-from-browser exposes active session credentials from the browser and should be treated as a last resort with explicit user consent.