Skip to content

generate-video-script

Use when generating a video script from a content brief. LLM-driven with configurable scene types, B-roll annotations, and hook/body/CTA structure. Supports multiple video styles and brand voice customization.

ModelSource
sonnetpack: video-pipeline
Full Reference

Generates a scene-marked video script from brief.json. Configurable video style, duration, hook technique, and brand voice. Every script follows Hook → Body → CTA structure with inline B-roll and graphic annotations. Output is script.md, consumed by build-storyboard.


ItemValue
Inputbrief.json in current directory
Outputscript.md in current directory
Next stagebuild-storyboard
Default styletalking_head
Default hookpattern_interrupt
Speech rate~130 wpm for runtime estimation

I want to…File
Configure video style, duration, hook type, scene types, and brand voicereference/configuration.md
See the script.md format, B-roll annotation syntax, and summary outputreference/output.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.


┏━ ⚡ generate-video-script ━━━━━━━━━━━━━━━━━━━━━━┓ ┃ Generating script: [title] ([style] / [tone]) ┃ ┗━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┛