Explainroo: Your AI Agent Can Now Make Fully Narrated Explainer Videos from a Prompt

Explainroo, an open-source tool by developer vincentsch, lets a coding agent turn a plain-text prompt into a fully narrated, animated explainer video MP4 rendered entirely on the user's own machine. The pipeline uses the Kokoro voice model for narration in 8 languages with 54 voices, OpenAI's Whisper for word-level timing, headless Chrome drawing frames in JavaScript with Rough.js and 1,800 Lucide icons, and a virtual camera with spring animations and motion blur; the agent writes only script.md and scenes.js. The project states it works with Claude Code, Codex, Pi and other coding agents, and currently works best with Claude Code on the strongest models.

Explainroo: Your AI Agent Can Now Make Fully Narrated Explainer Videos from a Prompt A post from RammCodes https://x.com/RammCodes has been making the rounds, and it points at something genuinely new: your coding agent can now produce a finished explainer video. Not a storyboard, not a script — a narrated, animated MP4. The tool is Explainroo https://github.com/vincentsch/explainroo by vincentsch, and it’s free and open source. We’ve covered a lot of agent-native tools here. This one stands out because video has, until now, been the one medium that stubbornly resisted the “describe it and get a result” workflow. What It Is Explainroo turns a plain-text prompt into a short explainer video where a voice narrates a topic while drawings appear in sync with the words. The whole pipeline runs on your own machine : - Voice — Kokoro https://huggingface.co/hexgrad/Kokoro-82M , an open voice model, reads the script in English, Spanish, French, Italian, Brazilian Portuguese, Hindi, Japanese, or Mandarin. 54 voices, no account, no API key. - Timing — Whisper https://github.com/openai/whisper listens to the narration and notes exactly when each word is spoken. - Pictures — Chrome runs headless in the background and draws the frames in JavaScript. Lines can look hand-drawn Rough.js https://roughjs.com , there are 1,800 Lucide https://lucide.dev icons, and scenes can show charts, math formulas, step-by-step code, your own screenshots, and video clips. - Camera — a virtual camera films each scene, pushing in on what the voice is describing and pulling back at the end, with little spring animations and motion blur, like a film camera. - Sound — it even generates its own background music per video. The Interesting Part: It’s Built for Agents Most video tools want you in an editor. Explainroo wants your coding agent in a terminal. The intended workflow is almost absurdly simple: Make me a short explainer video about your topic . Use explainroo for it: clone https://github.com/vincentsch/explainroo , read its AGENTS.md and follow the steps. The agent sets up Explainroo, writes two files, and renders the result: 1. script.md — the words the voice says. 2. scenes.js — the drawings, in a bit of JavaScript, each keyed to appear on a specific word from the script. Explainroo does everything else. You get back an MP4. The project notes it works with Claude Code, Codex, Pi, and other coding agents, and currently works best with Claude Code on the strongest models. Why This Matters Explainer video has been the expensive corner of content: you need a script, a voiceover, motion graphics, and someone who can sync all three. That’s why most documentation stays as text and most products ship without a demo video. Explainroo collapses that into a prompt. The use cases write themselves: - Product demos — generate a fresh walkthrough every release, in eight languages, without a studio. - Documentation — turn a tricky concept into a 60-second animated explainer as part of your docs build. - Courses and training — produce lesson videos at the pace you write lessons. - YouTube / marketing — explainer content without a motion-design bottleneck. Because the drawing layer is just JavaScript on a canvas, it’s inspectable and customizable — the same reason developers trust it that makes designers wary of it. This isn’t going to replace a bespoke, hand-crafted brand film. It’s going to replace the hundred explainer videos that never got made because the effort wasn’t worth it. The Catch It’s young, and the quality ceiling depends on the agent driving it and the clarity of your scene descriptions. A hand-drawn-canvas aesthetic is a style, not a universal fit. And “an agent made a video” still needs a human to watch it before it ships — timing glitches, awkward phrasing, and off-topic visuals are exactly the kind of thing that slips through automated generation. The Takeaway Explainroo is the first tool we’ve seen that makes “describe a topic, get a narrated animated video” actually work — locally, for free, and driven by the coding agent you already use. If you ship products, docs, or courses, this is worth a weekend: point your agent at the repo and ask it to explain something. Explore it: github.com/vincentsch/explainroo https://github.com/vincentsch/explainroo · explainroo.com https://www.explainroo.com .