gitwtfhub

wtf is vox-director?

alisa0808/vox-director — explained in plain English

Analysis updated 2026-05-18

349PythonAudience · vibe coderComplexity · 3/5LicenseSetup · moderate

TL;DR

An agent skill that turns a one-line topic into a finished Vox-style paper-collage explainer video, fully automated except for two approval steps.

Mindmap

mindmap
  root((Vox Director))
    What it does
      Writes narrative beats
      Generates collage posters
      Animates and narrates video
    Tech stack
      Python
      ffmpeg
      Atlas Cloud API
    Pipeline
      Beat map approval
      Style bake off
      Keyframes and motion
      Voice and music
    Use cases
      Explainer videos
      Ad style short films
      Educational content
    Audience
      Content creators

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

Why would anyone build with this?

REASON 1

Turn a one-line topic into a finished paper-collage style explainer video automatically.

REASON 2

Produce a short ad style video with narration, music, and captions with minimal manual editing.

REASON 3

Generate several visual styles for the same opening beat and pick the one you like before committing.

REASON 4

Animate real people or brand content using the frame-by-frame local motion path instead of a general AI video model.

What's in the stack?

PythonffmpegAtlas Cloud APIPillow

How it stacks up

alisa0808/vox-directorzli12321/lhtbkuskhan/jetendard
Stars349345344
LanguagePythonPythonPython
Setup difficultymoderatehard
Complexity3/54/52/5
Audiencevibe coderresearcherdeveloper

Figures from each repo's GitHub metadata at analysis time.

How do you spin it up?

Difficulty · moderate Time to first run · 30min

Requires a Claude Code or similar coding agent, an Atlas Cloud API key, and local ffmpeg plus Python with Pillow installed.

Free to use, modify, and distribute for any purpose, including commercial use, as long as the original copyright notice is kept.

Wtf does this do

Vox Director turns a single topic you type into a finished short explainer video, made in the paper collage style popularized by Vox's explainer videos, with hand cut looking paper cutouts, torn edges, tape, and bold headline text, all animated with narration, music, and captions. It is built as an agent skill, meaning it is not a standalone app but a workflow that a coding agent like Claude Code or Codex can follow, calling the Atlas Cloud API for AI models and the ffmpeg tool for video editing on your own machine. The process runs through several automated stages driven by a single project file. First the agent drafts a narrative arc for your topic and writes out a beat map, which you approve before anything else happens. Second, it renders the same opening beat in three or four different visual styles so you can pick the look you want by eye. From there it generates one finished collage poster image per beat, then animates each poster into a moving clip using an AI video model. A narrator voice and background music are generated separately, and everything is assembled into a final video with ffmpeg, which lines up the music under the narration and burns in captions and a watermark. For content involving real people or brands, there is a different animation path that avoids some of the content restrictions of general purpose AI video models by cutting the poster into individual pieces locally and animating them frame by frame instead of asking an AI model to animate the whole image at once. To use it, you install the skill files into your coding agent's skills folder, set an Atlas Cloud API key, and make sure ffmpeg and Python with the Pillow image library are installed. You then simply describe the video you want to your agent, for example a short collage video introducing a topic in a chosen language and aspect ratio, and the agent runs the whole pipeline, stopping twice to ask which beat map and style you prefer before producing the finished video file. The project is released under the MIT license.

Yoink these prompts

Prompt 1
Install the Vox Director skill and ask my coding agent to make a 30 second collage video about a topic I choose.
Prompt 2
Explain what happens at the beat map and style bake-off approval gates in Vox Director's pipeline.
Prompt 3
Show me how to set my ATLASCLOUD_API_KEY and install ffmpeg and Pillow for Vox Director.
Prompt 4
Walk me through the difference between the living poster animation path and the local frame-by-frame engine.

Frequently asked questions

wtf is vox-director?

An agent skill that turns a one-line topic into a finished Vox-style paper-collage explainer video, fully automated except for two approval steps.

What language is vox-director written in?

Mainly Python. The stack also includes Python, ffmpeg, Atlas Cloud API.

What license does vox-director use?

Free to use, modify, and distribute for any purpose, including commercial use, as long as the original copyright notice is kept.

How hard is vox-director to set up?

Setup difficulty is rated moderate, with roughly 30min to a first successful run.

Who is vox-director for?

Mainly vibe coder.

View the repo → Decode another repo

This repo across BitVibe Labs

Don't trust strangers blindly. Verify against the repo.