Repolode
Alisa0808

Alisa0808/vox-director

An agent skill that turns one topic into a finished Vox-style paper-collage explainer video, automating script, images, motion, voice-over, music and captions.

Stars
2,668
Forks
370
Open issues
6
Last commit
last month

What is it?

Vox Director is an agent skill that produces a Vox-style paper-collage explainer or ad video from a one-line topic. It runs the whole pipeline: a beat map (script), collage keyframe posters, animated motion, a narrator voice-over, background music, and burned-in captions, with the final output assembled as an mp4 using local ffmpeg.

The generation steps call the Atlas Cloud API for image, video, speech and music models, and the skill includes two human checkpoints: approving the beat map and choosing a visual style. It also supports two other inputs: re-styling an existing talking-head video into the collage look, and placing a single photo or presenter image inside a generated collage scene.

The skill works with any coding agent that can follow a workflow and run scripts, such as Claude Code or Codex. It requires an Atlas Cloud API key, ffmpeg and ffprobe, and Python 3 with Pillow.

Key features

  • 01Generates a full video from one topic: script, posters, motion, voice-over, music and captions.
  • 02Produces collage-style keyframes with torn paper, cut-outs and halftone looks.
  • 03Asks the user to approve the beat map and choose a style before generating.
  • 04Supports re-styling an existing talking-head video into the collage look.
  • 05Assembles the final mp4 locally with ffmpeg, including music ducking and watermark.

Use cases

  • Creating short explainer videos on a history, food or business topic without manual editing.
  • Producing product or brand ads in an editorial paper-collage style.
  • Turning a talking-head recording into a stylized collage video.
  • Prototyping explainer video concepts quickly with an AI coding agent.

Quick start

  1. 1

    Install the skill by cloning this repo into your coding agent's skills folder

    git clone https://github.com/Alisa0808/vox-director.git ~/.claude/skills/vox-director
  2. 2

    Set your Atlas Cloud API key in the environment

    export ATLASCLOUD_API_KEY="sk-..."
  3. 3

    Install ffmpeg and Python's Pillow library for caption and watermark overlays

    brew install ffmpeg && pip install pillow
  4. 4

    Ask your coding agent with the skill installed to make a video from a topic, then approve the beat map and style

Things to know

  • Requires a paid Atlas Cloud API key to generate images, video, voice and music.
  • Needs ffmpeg and Python installed locally, and the workflow depends on several external AI models.
  • The skill is only a workflow for a coding agent such as Claude Code or Codex, not a standalone app.

In collections

Topics

FAQ

What is vox-director?
An agent skill that turns one topic into a finished Vox-style paper-collage explainer video, automating script, images, motion, voice-over, music and captions. Vox Director is an agent skill that produces a Vox-style paper-collage explainer or ad video from a one-line topic. It runs the whole pipeline: a beat map (script), collage keyframe posters, animated motion, a narrator voice-over, background music, and burned-in captions, with the final output assembled as an mp4 using local ffmpeg.
Is vox-director free to use? What license does it have?
vox-director is open source under the MIT license. Check the license terms before using it in commercial products.
Is vox-director still maintained?
Updated within 6 months. The last commit to vox-director was on Oct 5, 2026.
What can I use vox-director for?
Creating short explainer videos on a history, food or business topic without manual editing. Producing product or brand ads in an editorial paper-collage style. Turning a talking-head recording into a stylized collage video. Prototyping explainer video concepts quickly with an AI coding agent.
How do I install vox-director?
Install the skill by cloning this repo into your coding agent's skills folder: git clone https://github.com/Alisa0808/vox-director.git ~/.claude/skills/vox-director Set your Atlas Cloud API key in the environment: export ATLASCLOUD_API_KEY="sk-..." Install ffmpeg and Python's Pillow library for caption and watermark overlays: brew install ffmpeg && pip install pillow Ask your coding agent with the skill installed to make a video from a topic, then approve the beat map and style
Where can I get vox-director?
The source code and installation instructions are on GitHub: https://github.com/Alisa0808/vox-director

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects