Monster Tamer 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 Monster Tamer — A real run of this skill. The input is a synthetic portrait, not a customer photo.

Ready to use

Monster Tamer

A vertical clip with your own little creature

Use this skill

Your photo: Required
Output: 5-15s clip with sound, landscape 3:2
Category: freeform
Cost: 5 credits

video
creature
anime
vertical

What it does

What you get back

A vertical clip made to be posted. Your doodles fix who the person is, then it invents an original little creature and films a brand-new beat in three parts: the call, the creature appearing in a puff of sparkle, and the two of them squared up together on the last frame. The creature is built to real design discipline — a silhouette that still reads at coin size, one clear signature feature, and its own limited high-identity palette so it looks designed rather than generic. Bouncy squash-and-stretch on its landing, flat cel shading throughout. Portrait 9:16 at fifteen seconds, one credit per second; it comes to life in the background and lands in the chat.

When the agent picks it

Use when the user wants a vertical, post-ready clip of their doodle character with a little creature companion — "make me a monster tamer video", "my doodle with a cute creature", "call out a pet monster", "creature collector clip for a reel". Fifteen seconds, portrait 9:16, built for a phone feed. The creature is ORIGINAL and invented for the clip; creature-collector STYLE through craft only, never any named franchise, creature or character. Not for gently animating one exact picture: that is doodle-motion.

The full skill file

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

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

Monster Tamer

Film a NEW vertical clip starring a character the user already has doodles of, alongside a creature invented for this clip. The references define the person — hair, face, colours, outfit — not the composition, and never the creature.

The creature must be original, and that is a design brief not a disclaimer

Every recognisable creature-collector franchise rests on the same discipline, and the discipline is reproducible without borrowing a single character:

a silhouette readable at coin size — if it only works large, it is not a mascot
one clear signature feature rather than five competing ones
a limited, high-identity palette of its own, so it reads as a designed character instead of a colourful blob
rounded shape language, because rounded reads as friendly and triangular reads as threatening

Name no franchise, creature or character anywhere in the prompt. A named-creature request is an IP problem and one the model frequently refuses, which would look like our bug rather than a policy.

Vertical and fifteen seconds, on purpose

reference mode, because H3 Max ignores aspect_ratio in image mode and a portrait request from a square doodle returns square. Fifteen seconds because call → appear → bounce → settle together is four moments; five seconds fits one.