vox-editorial

skill
Security Audit
Pass
Health Pass
  • License — License: NOASSERTION
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Community trust — 10 GitHub stars
Code Pass
  • Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Pass
  • Permissions — No dangerous permissions requested

No AI report is available for this listing yet.

SUMMARY

A Claude Code skill that turns one researched topic into a finished vertical explainer film: cut-paper artwork animated across depth planes, free local narration, assembled with ffmpeg.

README.md

vox

Two Claude Code skills for turning one researched topic into a finished
narrated film. Same editorial register, two different ways of building it.

vox-skill vox-editorial
Look Paper collage, halftone cut-outs, offset red marker stroke Cut paper, saturated grounds, scissored subjects
Built with Remotion — React that renders video Python + ffmpeg
Frame 1920×1080, or 1080×1920 on request 1080×1920
Charts, documents, newspapers Drawn in code, animate natively Drawn locally with PIL
Sprites Public-domain images, lifted locally with Vision Generated, one subject per sprite
Cost Free. Narration, music and every rebuild are local. Sprite generation only

Both are editorial journalism, not advertising: they earn the next five seconds
by being genuinely interesting rather than by shouting. Both are free to run —
narration comes from Edge neural TTS and everything else happens on your machine.


skills/vox-skill — Remotion

the founding of Baylor University

Full film: baylor-founding.mp4 · 65s · 1920×1080

Everything in that frame except the three portraits is drawn in code: the state
outline, the flag, the charter, the wax seals, the fifteen-card grid, the stamps,
the timeline. The portraits are public-domain scans lifted with the macOS Vision
framework and screened into halftone locally.

stills

The thing that makes it work is that the narration is the clock. The
voiceover is generated first, and every scene length and every animation cue is
derived from its word boundaries — so a scene starts when a word is spoken, not
when a number in a file says so. Change the voice and the whole film re-times
itself with one command.

It also ships a music generator. A real documentary cue was measured to find what
the genre actually is, spectrally, and the bed is synthesised to match and then
sidechain-ducked under the narration.

Read the skill →


skills/vox-editorial — ffmpeg

a film about JC Builds

Full film: jc-builds.mp4 · 49s

Generates elements, not posters. The image model only ever draws subjects,
spaced out on a keyable field; the ground, the tape, the newsprint and the
headline are drawn locally and composed across depth planes. Depth then drives
parallax, drift and draw order at once, which is what the eye reads as space.

Tokyo

Read the skill →


Install

Copy the one you want into your skills directory:

git clone https://github.com/jaredcassoutt/vox-editorial.git
cp -R vox-editorial/skills/vox-skill ~/.claude/skills/
# or
cp -R vox-editorial/skills/vox-editorial ~/.claude/skills/

Layout note. This repo used to hold vox-editorial at the root. It now
holds both skills under skills/. If you installed before, re-copy from the
new path.

Requirements

Both need ffmpeg and Python 3 with numpy, pillow and edge-tts.
vox-skill also needs Node and, for local subject cut-outs, macOS with Xcode
command line tools.

Licence

MIT. See LICENSE and NOTICE.

Reviews (0)

No results found