AnyDownload Duck Cloud YouTube workflow
🔄 yt-dlp workflow · metadata first · stream later

YouTube prototype

Inspect a YouTube URL, then choose the download type you want.

This page now matches your AnyDownload direction. Instead of only fetching auto captions, the UI demonstrates a local-first workflow for MP4, MP3, subtitles, captions, transcripts, and thumbnails powered by yt-dlp metadata.

No database is assumed here. The page only simulates the metadata-first frontend flow you described.
✨ Ready! Paste a YouTube URL and click “Inspect media” to preview AnyDownload actions.

Reference snippet

Starter API outline for a Next.js or Vercel-style route.

import youtubedl from 'yt-dlp-exec';

export default async function handler(req, res) {
  const { url, mode, format } = JSON.parse(req.body || '{}');

  try {
    if (mode === 'inspect') {
      const info = await youtubedl(url, {
        dumpSingleJson: true,
        skipDownload: true,
      });

      return res.status(200).json({
        title: info.title,
        thumbnail: info.thumbnail,
        formats: info.formats,
        subtitles: info.subtitles,
        automaticCaptions: info.automatic_captions,
      });
    }

    // Example follow-up branch:
    // choose a format or subtitle mode and stream the output directly.
    return res.status(200).json({ ok: true });
  } catch (error) {
    return res.status(500).json({ error: 'Failed to process download request.' });
  }
}

Deployment notes

Important constraints captured on the page.

  • Serverless time and memory limits can break large downloads.
  • Streaming is better than save-then-send for most media responses.
  • A bundled yt-dlp binary strategy is safer than assuming Python exists in deployment.
  • The product direction is local-first, so files stay with the user instead of living in a database.