If you want to make a narrated explainer or product demo, Claude Code can use your written request to guide video production. Claude Code is Anthropic’s coding assistant for the terminal: it reads your instructions, writes and runs code, and manages files. Claude Code coordinates the work, while a separate tool creates the video frames and combines them with audio.
For narrated explainers and product demos, explainroo is the best fit: it is a free, open-source kit built to work with Claude Code and other coding agents. You describe the video, and the agent uses explainroo to create and check an MP4 on your computer. The steps below cover that prompt-based workflow, the files Claude Code creates, and the settings you can adjust.
Ask Claude Code to make the video with explainroo
Start Claude Code in a working folder and give it a specific topic. For example, name the audience and the main point you want the video to communicate. Then paste this prompt, replacing the bracketed text:
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 prompt tells Claude Code to clone explainroo, read its instructions, and follow its video-making steps. The agent sets up the kit, creates and checks the video, then gives you an MP4 file. Claude Code can run terminal commands and manage project files, while explainroo handles the video production.
For a product demo, tell Claude Code the product’s name and where to find its code or website. explainroo can recreate the product’s screens with the same colors, fonts, and button labels. It can then animate a pointer as it clicks buttons and enters text while the narration explains each action.
How explainroo turns a prompt into a video
Claude Code writes two files: script.md, which contains the spoken narration, and scenes.js, JavaScript that draws the video’s scenes. You can set a scene to appear when a chosen word is spoken, so the visuals change in sync with the narration.
explainroo uses Kokoro, an open voice model, to read the script aloud. It offers 28 English voices, with American and British options, and needs no account or API key for built-in narration. Whisper, a speech-recognition tool, analyzes the recording and marks when each word is spoken. That timing lets explainroo match each visual to the spoken words.
Chrome runs in the background and draws video frames on an HTML canvas, the browser’s drawing surface. Scenes can include charts, code, screenshots, and icons from Lucide’s library. For a hand-drawn appearance, explainroo supports Rough.js. It also creates background music and small sound effects, then lowers the music while the voice speaks. FFmpeg, a command-line tool for processing media, combines the video and audio into an MP4.
Review the scenes and adjust the result
Claude Code cannot watch the finished video. explainroo gives it still images of each scene, a contact sheet showing small frames together, a layout check for cut-off or overlapping text, and a speech check for words the voice pronounced incorrectly. These checks help the agent find visible problems, but you should still open and watch the MP4 before sharing it. Check that the narration is accurate, the pacing works, and the visuals support what the voice says.
To change the video, ask Claude Code for a specific revision, such as shortening the opening or simplifying a scene. The agent updates script.md or scenes.js and runs video production again. You can also select one of six built-in visual styles: paper, clean, chalk, blueprint, midnight, or editorial. The style is set in video.json.
Choose a format based on where viewers will watch. explainroo supports 16:9 wide video for YouTube and 9:16 tall video for Shorts, TikTok, and Reels. It also supports 4:5 video for Instagram and LinkedIn posts, plus square video. Tall, 4:5, and square videos include captions that highlight words as they are spoken. For those vertical formats, explainroo keeps text clear of the areas where platform buttons appear. The pace setting in video.json changes narration speed and pauses. It also changes animation speed.
Run explainroo yourself when needed
Claude Code can set up explainroo as part of the prompt workflow. For manual setup, you need Node.js, FFmpeg, and Chrome or Chromium. Run these commands:
git clone https://github.com/vincentsch/explainroo.git
cd explainroo
npm install
node bin/explainroo.js doctor --fetch
The first setup downloads the voice and timing models. explainroo does not require a graphics card. explainroo is developed on Linux, and its tests run on Linux, macOS, and Windows. The rendered file is saved at videos/<name>/out/video.mp4. See the explainroo documentation for project instructions, and browse example videos to see the available styles and formats.
The kit is free and open source under the MIT license. Optional AI illustrations through OpenRouter are paid per image; the other explainroo video steps cost nothing per video with built-in voices. Claude Code is a separate service with its own pricing.
Know what explainroo can and cannot make
explainroo is designed for narrated explainers and product demos built from drawings, charts, code, and screenshots. It does not edit existing camera footage, create talking avatars, generate live-action video, or provide a drag-and-drop editor. You describe changes to Claude Code instead.
The narration is in English and seven other languages. Check names, technical terms, and factual claims in the final video, and make sure you have permission to use any screenshots or other material you provide.