How explainroo makes a video

A coding agent made this video with explainroo. It didn’t use a video model, an API key or a video editor. Everything you see is drawn with JavaScript.

explainroo is a free, open source kit for coding agents. The agent writes two files. The first one is the script for the voice. The second one is a bit of JavaScript that draws the pictures.

A voice model called Kokoro reads the script. It runs on your own computer. Then Whisper listens to the recording and writes down the time of each word.

This way a picture can appear at the moment the voice mentions it. The agent writes one line of code, and when the voice says database, the database appears.

Chrome draws each frame in the background. There are five looks to pick from, and explainroo also makes the music and the sound effects.

The agent can’t watch the video, so explainroo turns it into things the agent can read. It makes still pictures of each scene and contact sheets that show a scene over time. A check also finds text that is cut off, bad timing and words the voice got wrong.

One command turns all of it into an MP4 file.

explainroo is free and open source. The code and these example videos are on GitHub.