Analysis updated 2026-05-18
Turn a single topic prompt into a finished Vox-style explainer video.
Generate a scripted, narrated, animated short video without manual video editing.
Produce consistent-looking keyframes across a whole video using a style anchor image.
Estimate and budget the AI generation cost before rendering an expensive video.
| ck42bb/vox-explainer-skill | adityagahlot/obsidian-desktop-widget | adtyahac/acrobat-editor | |
|---|---|---|---|
| Stars | 53 | 53 | 53 |
| Language | — | JavaScript | C++ |
| Setup difficulty | moderate | easy | moderate |
| Complexity | 3/5 | 2/5 | 2/5 |
| Audience | vibe coder | general | general |
Figures from each repo's GitHub metadata at analysis time.
Needs an Atlas Cloud API key and a local ffmpeg install with libass.
vox-explainer is a skill for Claude Code, the AI coding assistant, that turns a single topic you give it into a complete short explainer video in the style of Vox's animated videos. You give it a topic, and it handles every step needed to produce a finished video file: writing a script, recording a spoken voiceover, generating still keyframe images, animating those images, adding background music, and stitching everything together into a final video on your own computer. The process starts by writing the script broken into beats, or short segments, and then generating and timing the voiceover narration first. The length of the spoken narration sets the pacing for the rest of the video, so the visuals are built to match the speech rather than the speech being trimmed to fit preset clip lengths. A single approved starting image is used as a style anchor, and every other keyframe is generated as an edit of that same image, which keeps the color palette, textures, and framing consistent throughout the video instead of looking like a set of disconnected pictures. The animation added to each keyframe is intentionally subtle and follows a specific set of allowed and forbidden types of motion, so it moves more like professional motion graphics than like raw video footage. All of the AI generation, the script writing, the voice synthesis, the image and animation generation, and the music, goes through a single external service called Atlas Cloud, which needs one API key. The final step of combining everything into a video file happens locally on your machine using ffmpeg, a widely used video processing tool. Every project's progress is saved to files as it goes, so if one segment of the video needs changes, you can redo just that piece without starting the whole project over. Making a sixty second video this way costs roughly three to eight dollars at current pricing, with the animation step being the most expensive part, and the skill tells you the expected cost before it starts the pricier stages. The project is released under the MIT license, so it is free to use, modify, and share.
A Claude Code skill that turns a single topic into a complete Vox-style explainer video, handling script, voiceover, keyframes, animation, music, and final assembly.
Free to use, modify, and distribute, including commercially, as long as you keep the copyright notice.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly vibe coder.
This repo across BitVibe Labs
Verify against the repo before relying on details.