Open source54 · GitHub

Explainer Video for your AI agent

Ask for a video inside Codex, Claude Code, Cursor or any MCP client and get a narrated, hand-drawn explainer MP4 back. Text, a URL or a document in; storyboard, illustrations, voice-over and subtitles out. No API keys.

Try in browser

Install for Codex & Claude Code

terminal
$ npx --yes github:SpeedPainterOrg/explainer-video

Remote MCP server

prompt
https://api.speedpainter.org/mcp

Say it, get the video

› Create a 30-second explainer video about the history of the FIFA World Cup, from the first tournament in 1930 to today.

The agent plans a storyboard, draws each scene, narrates it and returns the MP4.

Real output · 40 s · one prompt

Installation

  1. 1.

    Open your terminal and run the installer. It detects Claude Code and sets up the skill and the MCP server.

    terminal
    $ npx --yes github:SpeedPainterOrg/explainer-video
  2. 2.

    Start a new Claude Code session. Already-open sessions keep an older tool snapshot.

  3. 3.

    If sign-in does not open by itself, run /mcp inside the session and choose explainer-video.

  4. 4.

    Ask for a video, for example: “Create a 30-second explainer video about the history of the FIFA World Cup.”

First use opens a Google sign-in. Credits are shared with your SpeedPainter account.Claude Code MCP documentation

What your agent can do

  • Text → video

    Paste notes, a script or a topic. 5 seconds to 5 minutes, 16:9, 9:16, 1:1 or 4:5, narration in the language you ask for.

  • URL → video

    Hand over a link. The agent reads the page with its own tools and sends only the extracted text.

  • Document → video

    Attach a PDF or notes. The file stays on your machine; you get the MP4, the subtitle file and the real duration back.

Only extracted text is sent; files and your API keys never leave your machine.No agent? Make one in your browserSource on GitHub