create-supercut

Use when creating a supercut or compilation video — analyzing transcripts to extract highlight moments, selecting clips by keyword or topic, assembling a timeline with transitions, getting user approval before rendering, and integrating background music.

Model	Source
sonnet	pack: video

Full Reference

create-supercut

A supercut is a curated compilation of clips around a theme, keyword, or emotional arc. The workflow is always: analyze transcripts → select and score clips → assemble timeline → get explicit user approval → render with optional music. Never render without approval.

Quick Reference

Item	Value
Transcript formats	JSON (Whisper/Deepgram/AssemblyAI), SRT
Clip padding default	2 seconds before/after keyword hit
Video component (SSR)	<OffthreadVideo> — never <Video>
startFrom / endAt units	Frames (multiply seconds × fps)
Approval required	Yes — present timeline before every render
Background music volume (with dialog)	0.3–0.5 max

Reference Index

I want to…	File
Parse SRT or JSON transcripts into segments	transcript-analysis.md
Find keyword hits, merge overlapping clips, or score by quality	clip-selection.md
Build the Remotion Sequence timeline, add fade transitions, or compute durationInFrames	timeline-assembly.md
Present the approval prompt, add background music, or run the final render	music-and-render.md

Usage: Read the reference file matching your current task from the index above. Each file is self-contained with code examples and inline gotchas.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Announcement

┏━ 🎬 create-supercut ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓ ┃ [one-line description of the supercut you’re building] ┃ ┗━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┛

Output this box as the first response before any work begins. No exceptions.