Doodle in Motion example output generated by Doodle AI — A real run of this skill. The input is a synthetic portrait, not a customer photo.

The photo used as input for Doodle in Motion — A real run of this skill. The input is a synthetic portrait, not a customer photo.

Ready to use

Doodle in Motion

Make the doodle you just made start moving

Use this skill

Your photo: Required
Output: 5-15s clip with sound, square 1:1
Category: freeform
Cost: 5 credits

video
animation
motion

What it does

What you get back

Takes a single finished doodle and animates it without redrawing it: the same marker line, the same flat colours, the same face. Expect small, springy, hand-drawn motion — a blink, a smile, a wave, hair drifting — with a near-still camera and a soft sound bed. Costs one credit per second of clip, and it renders in the background, so the clip arrives in the chat about a minute after you ask.

When the agent picks it

Use when the user wants an image they already have to start moving — "animate this", "make it move", "turn this doodle into a video". The picture they point at becomes the first frame of the clip, so the result is recognisably that exact drawing, animated. Not for a new scene starring the character: that is doodle-reel.

The full skill file

Every skill is one SKILL.md — this is the exact file the agent reads.

src/mastra/skills/doodle-motion/SKILL.md

Doodle in Motion

Animate ONE image the user already has. That image becomes the clip’s first frame, so the output is that drawing moving — not a new interpretation of it.

When this is the right skill

Pick this when the user points at something specific and asks for movement:

“animate this” / “make this move” / “can you make it a video”
“make the doodle you just made wave”
“turn my avatar into a short clip”

Choose doodle-reel instead when the user wants a NEW shot starring the character (“a clip of her riding a scooter”, “make a video of this character in a coffee shop”). The test is simple: if the user’s request describes a scene that is not already in the picture, it is a reel, not motion.

If there is no image yet, make one first with the right image skill, show it, and only then offer to animate it.

What to animate

The frame is fixed, so what you are choosing is the MOTION:

one clear primary action (a wave, a blink-and-smile, a small nod, a laugh)
secondary drift in hair, scarf, earrings, clothing
a background that stays a calm flat wash — it should not compete

Keep it small. A doodle that tries to do three things in five seconds reads as a glitch, not as charm.

How to run it

Confirm you have the image URL to animate. Usually it is the doodle you generated a moment ago in this same conversation; otherwise it is the user’s attached photo. If you have neither, ask — do not call the tool.
Call generateVideo with skill: "motion", that URL as imageUrl, and a short description of the action if the user named one.
Pass seconds whenever the user names a length. If you omit it the clip is 5 seconds. It costs 1 credit per second, so never quietly pick 15.
The tool returns status: "queued" and a jobId. That means the clip is RENDERING, not ready. Say so in one line — roughly a minute, it will appear here on its own, it has sound — and then stop.

What not to do

Do NOT call the tool again for the same request while a clip is rendering, and do not call it to “check” on one. There is nothing to poll: the finished clip is delivered to the app by itself.
Do NOT claim the clip is ready, describe what it looks like, or offer a download link. You have not seen it.
Do NOT promise a resolution, a frame rate or a silent clip. Clips render at 480p and always carry an audio track.