Claude cannot natively generate a finished video file from a text prompt. Anthropic’s Claude models work with text, images, tool calls, and computer interactions, but current evidence does not show that they generate video. Claude also does not process raw video files directly like a video model. Tools such as claude-real-video turn video into still frames and a transcript for Claude to analyze. This adds video-processing tools around Claude; it does not give Claude video-generation capabilities.
Claude can help create a video by writing the script, planning scenes, writing code, and directing other software. For narrated explainers and product demos, explainroo is the best fit: it lets a coding agent such as Claude Code turn the plan into an MP4 rendered on your computer.
What Claude can and cannot do
Claude can write a voiceover script, plan scenes, describe visual changes, and write code to draw them. A coding agent can run the code, create files, render frames, and assemble the video.
Claude does not generate a complete sequence of moving images. Current lists of AI video generators include Runway, Sora, Pika, and Kling. They generally describe Claude as a language and multimodal assistant, not a video generator. Some video platforms use Claude to refine prompts or manage a conversation while a separate model creates the footage. Claude plans or controls the process, and that model renders the video.
Anthropic’s Claude Design has been described as a visual creation and prototyping platform that can include video in prototypes. This does not show that ordinary Claude conversations can generate video. Anthropic’s products and their capabilities change with new releases, so check its current documentation for product-specific updates.
The practical rule is simple:
- Ask Claude to write and plan the video.
- Use a rendering tool to draw, animate, record, or assemble it.
- Treat Claude as the director and coding assistant unless a separate video model handles image generation.
How explainroo turns Claude’s plan into a video
explainroo is a free, open-source kit released under the MIT license. An AI coding agent uses it to write scene code and render narrated explainers and product demos on your computer.
The agent creates two central files:
script.mdholds the narrator’s spoken words.scenes.jscontains JavaScript that draws each scene.
A scene can appear when the narrator reaches a chosen word, so the visuals follow the narration without rough, manually estimated timestamps.
explainroo runs the rest of the production steps on your computer:
- Kokoro, an open voice model, reads the script aloud using English voices. No account or API key is required for Kokoro, and narration is in English and seven other languages.
- Whisper listens to the recording and records when each word is spoken.
- Chrome or Chromium draws every frame on an HTML canvas.
- Rough.js can give lines a hand-drawn appearance, and Lucide provides icons.
- Scenes can include charts, code, and screenshots.
- explainroo creates background music and adds small sound effects. It lowers the music while the narrator speaks.
- FFmpeg combines the narration, visuals, and sound into an MP4 file.
explainroo renders the video on your computer, so you get a finished video instead of suggestions that need manual editing.
How to make a video with Claude and explainroo
Claude Code is the quickest option. explainroo also works with Codex, Pi, OpenCode, Gemini CLI, and other coding agents that can run shell commands. It works best with Claude Code and Opus.
Give the coding agent this prompt, replacing the brackets with your subject:
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, then writes the script and scene code. After rendering the video, it checks the result and returns the MP4 file. The file is saved at:
videos/<name>/out/video.mp4
For a product demo, include the product name and the location of its code or website. explainroo can use that material to rebuild the product’s screens with its colors and fonts, and reproduce the button labels. A generated pointer can click through screens and type in fields as the narration explains each step.
To install explainroo yourself, you need Node.js and FFmpeg. You also need Chrome or Chromium. You do not need a graphics card. explainroo is developed on Linux, and its tests run on Linux, macOS, and Windows.
The basic installation commands are:
git clone https://github.com/vincentsch/explainroo.git
cd explainroo
npm install
node bin/explainroo.js doctor --fetch
What kinds of videos Claude can help create
Claude and explainroo work especially well for videos built from narration and designed scenes. Examples include:
- Product walkthroughs
- Software tutorials
- Technical explainers
- Process diagrams
- Feature announcements
- Code demonstrations
- Educational videos
- Social media posts with narration
You can select one of six built-in visual styles in video.json: paper, clean, chalk, blueprint, midnight, or editorial. The same setting controls the overall look without requiring manual redesign of every scene.
explainroo supports several video shapes. Wide video suits YouTube. Tall video suits YouTube Shorts, TikTok, and Instagram Reels, with text placed away from areas covered by platform controls. It also supports portrait and square video for social media posts. These formats include captions that highlight each spoken word.
The pace setting in video.json changes the speed of the voice, pauses, and animations.
How explainroo checks the result
An AI coding agent cannot watch a finished video as a person would. explainroo gives the agent several kinds of feedback:
- Still images from each scene
- A contact sheet of small frames from the video
- A layout check that detects cut-off text and overlapping elements
- A speech check that identifies words the voice pronounced incorrectly
This feedback lets Claude revise the scene code and narration without watching the video. You should still watch the final MP4 yourself when visual accuracy, product behavior, or brand presentation matters.
What explainroo does not create
explainroo makes narrated videos and product demos from code-drawn scenes. It does not edit existing camera footage, create talking avatars, or generate AI live-action footage. It has no drag-and-drop editor. Instead, you tell the coding agent what to change, and it updates the script or scene code.
Optional AI illustrations are available through OpenRouter and are paid for per image. Everything else in explainroo costs nothing per video, while the coding agent remains a separate service with its own pricing.
explainroo is a poor fit for projects that need recorded interviews, cinematic live-action scenes, or an avatar speaking to the viewer. It works well for structured explainers with precise timing, diagrams, interface screens, captions, and narration.