How I use Todrawn Scribe and MCP to generate content

Most of the short videos I post for Todrawn look the same: something gets drawn by hand, one stroke at a time. Drawing each of them myself took too long, so I gave the job to my coding agent.

Todrawn Scribe has an MCP server. Any agent that supports MCP can build a Scribe project and give me a link. I open the link, watch the pen draw, and export the video if I like it.

Setup was one command in Claude Code. No API key and no config file, the server runs on my machine.

claude mcp add todrawn-scribe -- npx -y @todrawn/scribe-mcp

One photo of my dog

My prompt was short: paint dog.png like a person would.

A person doesn't paint a photo from top to bottom like a printer. They start with the ground, then the subject, then the background. So the agent split the photo into six parts and painted them in this order: the deck, the dog, the baskets, the table, the tree and the house. Then it sent all the strokes to the server in one call.

> paint dog.png like a person would

regions: deck -> dog -> baskets -> table -> tree -> house
scribe_create_drawing({ paths, viewBox: "0 0 540 960", aspect: "portrait", pen: "fineliner", background: "paper" })
Artboard: portrait, 3756 strokes, 17862 points. The whole drawing is inside the link after "#", so nothing was uploaded.

The pen replays the strokes in the same order the agent sent them. The full replay takes about three minutes, so in the clip I posted it is sped up to 48 seconds. I added the captions, music and voiceover afterwards.

You can watch the clip in the original post.

Nothing gets uploaded

The Scribe tools only build links. The whole drawing, all 3,756 strokes, is stored in the part of the link after the #, and browsers never send that part to a server. The photo never left my laptop.

What else the server can do

• scribe_create_video: you give it code or text and get a handwriting video, with syntax colours for code.
• scribe_create_drawing: the agent sends its own strokes as SVG paths and the pen draws them in that order. This is what I used for the dog.
• scribe_paint_image: paints a PNG or JPEG in one call, on your machine, with up to about 2,800 strokes. Good when you don't care about the order.
• Board tools: after signing in once, the agent can also put diagrams, notes and tables on your Todrawn whiteboards.

What I still do myself

• I pick the idea and write the prompt. The agent is good at drawing, not at knowing what people want to watch.
• I open the link, check the drawing and export it. The export runs in my browser and needs a free Todrawn account.
• I edit the video: captions, music and voiceover are done outside Scribe. Since I'm not a native speaker, I use ElevenLabs to generate the voice.

Try it

The MCP server page has install commands for Claude Code, Claude Desktop, Cursor, VS Code, Codex and Gemini CLI. Watching what the agent made is free and needs no account.