ShipCut

Desktop-first AI video repurposing pipeline.

Drop in a YouTube URL or a local .mp4, get:

• a cleaned transcript
• a condensed long-form cut targeting about 10 minutes
• vertical 30s and 60s shorts with burned captions
• publish bundles for YouTube, Instagram, and TikTok

Monorepo

shipcut mirrors the shipcode repo shape:

apps/
  cli/
  desktop/
  web/
packages/
  db/
  media/
  publish/
  shared/
  story/
  transcript/
  ui/
skills/

MVP pipeline

1. Ingest local video or download source media from YouTube
2. Reuse available subtitles when possible
3. Fall back to local transcription for raw media
4. Analyze narrative sections and highlight windows
5. Render one long-form cut plus 30s and 60s shorts
6. Generate platform metadata and publish bundles

Local prerequisites

• Bun
• ffmpeg
• ffprobe
• yt-dlp
• uv

The local transcription fallback installs faster-whisper on demand through uv.