vincentsch/explainroo

The video studio your AI agent runs.
Free and open source. The voice, the timing and the rendering run on your computer.

Website  ·  Example videos  ·  Docs  ·  AGENTS.md

https://github.com/user-attachments/assets/6dd5dc32-c975-4e0e-8cd4-2ef3ccc11e61

_{A coding agent made this video with explainroo. You can also watch it on explainroo.com.}

News

• Nov 10, 2026: explainroo 0.6 records your real app with explainroo record, and a scene plays that recording with a pointer and zooms that land on the narration. New projects start in the studio look, and formats give one kind of video a ready shape. check and verify work on any video file, and cut edits your own recording by its words. The list of sizes moved to explainroo sizes.
• Nov 9, 2026: explainroo 0.5 runs with npx, so you don't have to clone it. It can also use outside services on your own key, like AI video clips and posting to your channels. German narration works with a paid voice or your own recording. A project made before keeps its look, frame for frame, until you add "defaults": "0.5" to its video.json, which brings the new look, like bigger arrow labels.
• Oct 20, 2026: explainroo 0.4.1 keeps arrow and chart labels whole while a shot holds.
• Oct 15, 2026: explainroo 0.4 films each scene with a camera that follows the narration.
• Oct 13, 2026: explainroo 0.3 narrates in Japanese and Mandarin Chinese, and adds the editorial look.
• Oct 13, 2026: explainroo 0.2 adds five more languages, other voices and video clips.

All news is on explainroo.com/news.

Make a video

[!TIP] Tell your coding agent to use explainroo, and what the video should explain. It works with Claude Code, Codex, Pi and other coding agents. Right now it works best with Claude Code and Opus 5.5.

Copy this into your agent and put your topic in place of the brackets:

Make me a short explainer video about [your topic].
Use explainroo for it: run npx --yes explainroo docs and follow what it prints.

The agent sets up explainroo, makes the video and checks it. You get an MP4 file.

How does it work?

An explainer video is a short video where a voice explains a topic and drawings appear while it speaks. For explainroo, the agent writes two files. script.md has the words the voice says. scenes.js draws the pictures with a bit of JavaScript, and each drawing can appear on a word from the script. explainroo does the rest:

• Voice. Kokoro, an open voice model, reads the script aloud in English, Spanish, French, Italian, Brazilian Portuguese, Hindi, Japanese or Mandarin Chinese. It has 54 voices and needs no account or API key. You can also use an ElevenLabs or Cartesia voice, or record the narration yourself. German works with one of these, because Kokoro has no German voice. One project can make the same video in several languages, with a script for each.
• Timing. Whisper listens to the narration and notes when each word is spoken.
• Pictures. Chrome runs in the background and draws the frames. The lines can look hand drawn (Rough.js), and there are 1,800 icons from Lucide. A scene can also show charts, math formulas, code that changes step by step, your own screenshots and video clips such as screen recordings.
• Camera. A camera films each scene. It moves in on what the voice talks about when there is time, and pulls back at the end so the last frame shows the whole idea. Cards and labels are never cut in half while a shot holds. Things arrive with a little spring, and the final video has motion blur, like a film camera.
• Sound. explainroo makes its own background music for each video and adds small sound effects. The music gets quieter while the voice speaks.
• File. ffmpeg puts it all together into an MP4, with subtitle files, YouTube chapters and a cover image.

An agent can't watch a video, so explainroo gives it other ways to check its work. It saves stills of the scenes and a sheet of small frames for the whole video. A layout check finds text that is cut off or overlaps, and a speech check finds words the voice got wrong. For a feed-sized player, explainroo check videos/<name> --view-width 854 also flags small text.

Check or cut any video

check and verify also take a video file made with any tool:

npx --yes explainroo check talk.mp4 --script script.txt

They report:

• black or blank frames, and stretches where the picture stands still
• silence, the loudness and the true peak
• text too small to read on a phone, found in frames of the video
• with --script, the words of your script that the speech model didn't hear

--json gives the same report for your agent.

explainroo cut talk.mp4 edits your own recording by its words. It writes talk.cut.md, one sentence a line with the time of each word. Your agent deletes words and long pauses there. Then explainroo cut talk.mp4 --apply writes talk.cut.mp4, with a short crossfade at every join so the cuts don't click. Both run on your computer, like the rest of explainroo.

What leaves your computer

Making a video needs no account and no key, and it costs nothing. The voice, the timing, the music and the rendering run on your computer. Your coding agent is a separate service with its own terms and prices.

Some commands send something to an outside service, and only when you or your agent run them:

You run	What leaves your computer	Where it goes
connect	A test request with your key, or the code from your sign-in	That service
voice with ElevenLabs or Cartesia	The narration text	That company
voices --provider elevenlabs or cartesia	Your key, to list its voices	That company
image or clip	The prompt and any reference images	OpenRouter and the model's provider
clip --dry-run	A request for the list of video models and their prices, without your prompt or a key	OpenRouter
services, when it lists clips	The same request as clip --dry-run, at most once a day. It asks for the list of video models and their prices, without your prompt or a key	OpenRouter
share	Nothing in this version, because the hosted service is coming. Once it runs, the video file and its title	explainroo.com
publish	The video or a link to it, its title and its text. With a link, first a request for the size of the file there	YouTube, Buffer or Blotato, and whatever they send it to. The size request goes to the host of the link
publish --dry-run	Your login, to read your channels or accounts, and the same size request	YouTube, Buffer or Blotato, and the host of the link
capture	Requests for the page you name and the files it loads	That site and the servers it loads from
formats add --yes	A git clone of the address you name. Without --yes it only prints what it will run	The host of that Git repository
presenter	The narration of one scene, as sound, and the id of the avatar. --avatars and --dry-run send only your key	The presenter service you set up. None is built in yet

With ElevenLabs or Cartesia, the commands that make missing narration, like check and render, send the narration text too.

The first run downloads the voice and speech models from Hugging Face. Other languages need one more speech model, which downloads the first time, and Japanese and Chinese also download their fonts from GitHub. On Debian and Ubuntu, doctor --fetch downloads espeak-ng's packages from the computer's own package sources when a language needs it. Nothing of yours goes along with these downloads. explainroo has no telemetry, so it sends nothing about how you use it.

Nothing uploads to the hosted service without the explicit explainroo share command. In this version share sends nothing at all, because the service doesn't run yet.

Looks

There are seven looks: studio, paper, clean, chalk, blueprint, midnight and editorial. A new project starts in studio, which is light and calm, with white cards and lists on cards. You change a video's look with one setting in video.json. A project can also bring its own look, a small JSON file with its colors, fonts and background. Here is one frame in six of the built-in looks, from the example videos.

Sizes

You pick the size for the place the video goes. YouTube videos are wide, and Shorts, TikTok and Reels are tall. Instagram and LinkedIn posts use 4:5, and there is a square size too. Shorts, TikTok and Reels put their own buttons over the video, and explainroo keeps your text out of those spots.

Tall, 4:5 and square videos get captions that light up word by word. "captions": "bold" shows two to four big words at a time instead, with numbers and marked words in color. For Shorts, TikTok and Reels, s.short draws the parts short videos often have, like a hook title that is there on the first frame (explainroo docs short-video). One project can render several sizes: explainroo render videos/demo --size tiktok writes the TikTok version next to the wide one. "sizes" in video.json gives one size settings of its own, like bold captions and a quicker pace for TikTok only.

More than an MP4

Every render also writes subtitle files (.srt and .vtt) that YouTube and LinkedIn accept, a cover image for the thumbnail, and YouTube chapters when the script names them. --format gif makes a GIF for a README or a post. --transparent --format mov makes a video with a transparent background, for putting a diagram over camera footage in a video editor. explainroo export writes a small web player that draws the video sharp at any size and embeds with an iframe.

Pace

"pace": 1.2 in a video's video.json makes the voice, the pauses and the animations 20% quicker. The music gets a little quicker too. Pace goes from 0.7 to 1.6, and 1 is normal.

Product demos

explainroo can also make product demos, the videos software companies make to show their app. The agent rebuilds the app's screens from its code or its website, with the same colors, fonts and button labels. A mouse pointer then clicks through the screens and types into the fields. Tell your agent which product it is and where to find its code or website. explainroo capture saves the live site's screenshots, colors, font files, logo and wording first, with the page's text and a list of its images. A real screen recording can play inside a scene too.

Here are the demos for Unspar and Vroni, two of my own products. The files for the Unspar demo are in examples/unspar-demo. explainroo.com also has unofficial demos of Gmail, ChatGPT and Claude.

Formats

A format is a ready way of making one kind of video, with a short guide for your agent, scene helpers, settings and a whole example video. Four come with explainroo:

Format	For
product-launch	a launch video of about a minute
narrated-explainer	one idea explained in about a minute
vertical-short	a short for Shorts, TikTok and Reels, with the hook in the first second
product-ad	an ad of 15 to 30 seconds for a feed

explainroo init videos/acme --format product-launch starts from one. Anyone can publish a format as a Git repository, and explainroo formats add <address> installs it. It first prints the git commands it will run and installs only with --yes. A format's code runs only when you make a video from it, and its author is named in the credits of every package (explainroo docs formats).

Outside services

Each step of making a video has a free choice first. The services that cost money use your own account and key:

Step	Free	Costs money
Voice	Kokoro, or your own recording	ElevenLabs, Cartesia
Timing	Whisper
Images	The built-in icons and your own files	OpenRouter
Clips	Your own footage	OpenRouter
Presenters	Film yourself	An AI presenter service (coming, none built in yet)
Music	The built-in music, your own file, or Soundroo
Sharing	A web player for your own site, or a link from share (coming)
Posting	A package you upload yourself, YouTube, or Buffer on its free plan	Blotato

explainroo services lists every choice with what it costs and what data leaves your computer. For clips it also shows today's price per second. explainroo connect <service> sets one up and tests your key or login. One OpenRouter key makes images and clips. explainroo's instructions tell your agent to show you the options, free ones first, and to let you choose.

A project keeps a record of what was paid for in its history/ folder. The same request again uses the file it already bought. After a crash, a clip that was already ordered is picked up again, and a request whose outcome is unknown is never sent again by itself. explainroo history shows the record, and explainroo docs connectors says how to add a service.

Presenters

A presenter is a person who talks on screen next to the product. The free choice is to film yourself. explainroo presenter can also make a clip of an AI presenter who speaks the narration of a scene, through a presenter service you pick. The code and its rules are built, and no presenter service is built in yet. explainroo never makes a real person's face or voice without the service's consent flow, and every package of a video with an AI presenter gets the AI label (explainroo docs presenters).

AI video clips

explainroo clip buys a short video clip that an AI model makes from a text prompt. It goes through OpenRouter on your own key, and the clip plays in a scene like any other video file. --dry-run shows the price and spends nothing, also before you have a project, and it lists the lengths the model takes. A paid run needs --max-cost or a budget. Before it buys, clip checks the price against that cap and stops when the price is higher. The final cost is known only after the job, and it can come out a little higher than the price. "clipBudget": 0.40 in video.json, or --budget 0.40, is one budget for all clips of a video. With it --max-cost is optional, and clip stops before a clip that would go over it. Clips work for short shots of the real world, like a street or food. They get text and diagrams wrong, and a character looks different in each clip. explainroo docs clips has the details.

Music and sound files

A video can play your own music file instead of the built-in music. explainroo brings it to the level of the built-in music and lowers it while the voice speaks. s.sound plays a sound file on a word. Soundroo is a free tool where your agent writes original music, and it renders the track on your computer. explainroo docs music has the details.

Posting a video

explainroo package writes a folder with everything needed to post a video on one platform: YouTube, Shorts, TikTok, Reels, LinkedIn or X. Next to the video at the right size, it holds the words for the post. It also says whether to mark the post as made with AI, and why. You can upload the folder yourself.

explainroo share uploads a finished video to explainroo.com and gives back a page link. That hosted service is coming. Until it runs, share sends nothing, says so and makes no link.

explainroo publish posts a package through YouTube, Buffer or Blotato, with your own account. --dry-run shows the exact post first, and a real run posts only after a yes. explainroo docs publishing has the details.

Partners

Partners are coming. A partner will be a company that pays to be built in as one of the choices for a step. There are none yet. explainroo services always labels a partner as a "paid partner", and the free choices stay first in every step. explainroo's instructions tell your agent never to pick a paid service for you.

Install it yourself

npx --yes explainroo doctor --fetch

This checks your computer and downloads the voice and timing models once, about 400 MB together. You don't have to clone anything. You need Node.js 20.11 or newer, ffmpeg, and Chrome or Chromium. You don't need a graphics card. I develop explainroo on Linux, and its tests run on Linux, macOS and Windows for every change.

On Linux, installing explainroo also makes ONNX Runtime, which the voice uses, download its graphics card files from GitHub. explainroo doesn't use them. When GitHub is slow, that download fails and npx stops without a word. Put ONNXRUNTIME_NODE_INSTALL_CUDA=skip in front of the command to leave the download out:

ONNXRUNTIME_NODE_INSTALL_CUDA=skip npx --yes explainroo doctor --fetch

For narration with Kokoro, the free voice, in Spanish, French, Italian, Portuguese or Hindi, also install espeak-ng, for example with sudo apt install espeak-ng or brew install espeak-ng. npx --yes explainroo doctor --language es checks what Spanish needs. On Debian and Ubuntu without root, npx --yes explainroo doctor --fetch --language es puts a copy of espeak-ng in explainroo's cache folder instead. Japanese and Chinese need Python with misaki instead, as described in docs/japanese-and-chinese.md. German narration needs neither, and it works with ElevenLabs, Cartesia or your own recording, as docs/languages.md describes.

Then ask your agent for a video in any folder, for example "Make a 60 second video about how HTTPS keeps a password secret. Use explainroo: run npx --yes explainroo docs and follow what it prints." The agent saves the video in its project folder as out/video.mp4. The example videos are in examples/, and their sources come with the package too, in the folder that explainroo doctor prints. The full documentation is on explainroo.com.

Your agent can also keep explainroo as a plugin or skill, so it knows the tool without being told:

• Claude Code: /plugin marketplace add vincentsch/explainroo, then /plugin install explainroo@explainroo
• Codex: codex plugin marketplace add vincentsch/explainroo, then codex plugin add explainroo@explainroo
• skills.sh: npx skills add vincentsch/explainroo

To change the code, clone it instead:

git clone https://github.com/vincentsch/explainroo.git
cd explainroo
npm install
node bin/explainroo.js doctor --fetch

The watermark

Each video has a small "explainroo.com" in one corner. "watermark": false in video.json turns it off. Please keep it if you can, because that is how other people find explainroo. A TikTok package has no watermark, because TikTok's rules don't allow apps to add one. The publishing page quotes them.

Credits and license

explainroo is MIT licensed. The voice comes from Kokoro, and the word timing from Whisper through Transformers.js. The drawings use Rough.js and Lucide icons. MathJax typesets the formulas. Playwright runs Chrome, and ffmpeg makes the video file. For other languages, explainroo runs espeak-ng as a separate program, with phoneme rules from Kokoro's misaki, and misaki itself for Japanese and Chinese. OpenCC and pinyin-pro help the speech check compare Chinese, and HarfBuzz keeps web player fonts small. The fonts, including Noto Sans CJK for Japanese and Chinese, are under the SIL Open Font License.