ShipCut

Desktop-first AI video repurposing pipeline. Feed in a YouTube URL or local .mp4, resolve the transcript, condense the story, render a long-form cut, generate vertical shorts, and package publish-ready artifacts from one run.

What ShipCut does

Stage	What happens
Ingest	Pull source media from YouTube or read a local file
Transcript	Reuse source subtitles first, then fall back to local Whisper
Analysis	Score narrative sections, find highlight windows, and build edit plans
Render	Produce one 16:9 long-form cut plus 15s / 30s / 60s shorts
Publish	Write metadata bundles and upload handoff artifacts for each target

Why ShipCut?

Most video tools assume you want to live inside a full editor. ShipCut is built for the opposite case: large source footage stays local, transcript-driven decisions stay inspectable, and the output is already organized for YouTube, Instagram, TikTok, and X.

Transcript-first keeps cut decisions explainable instead of opaque
Desktop-first keeps heavy media on your machine instead of uploading it to a SaaS
One workspace per run keeps source, transcript, renders, and publish metadata together

Start here

Getting Started — prerequisites and local setup
CLI — process and list commands
Architecture — monorepo layout and package boundaries