Doodle Idea example output generated by Doodle AI
A real run of this skill. The input is a synthetic portrait, not a customer photo.
Ready to use

Doodle Idea

Draw the thing you have in mind — no photo needed

Use this skill
Your photo
Not needed
Output
Single square 1:1
Category
freeform
Cost
1 credit
  • freeform
  • no photo
  • ideas
  • objects
  • scenes

What it does

What you get back

Give it an idea in words and it draws that idea as a naive marker-and-ink doodle: a cat on a skateboard, a teapot garden, a tiny rocket, a bowl of ramen. With nothing typed it picks one interesting concrete subject to draw. Built for the "give me something to doodle" moment, not for portraits.

When the agent picks it

Use when the user wants a doodle of a THING they have in mind — an object, a scene, a motif, a creature, a small still life — drawn in the house doodle style, with no photo. Phrases like "draw a cat riding a bike", "doodle idea", "give me something to draw", "random doodle", or a bare noun. For a random FICTIONAL CHARACTER instead of the thing they typed, use surprise-me. For a doodle of a real person or pet from a photo, use doodle-avatar or pet-portrait.

The full skill file

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

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

Doodle Idea

Draw the idea the user typed — an object, a scene, a motif, a creature, or a small still life — as a single hand-drawn doodle in the house style. No photo required.

When this is the right skill

Pick this skill when the user gives you an idea for a thing to draw and wants it rendered as a doodle:

  • a concrete subject in words — “a cat riding a bicycle”, “a cactus in a mug”, “a little haunted house”, “a bowl of ramen”
  • an open-ended “give me a doodle idea”, “random doodle”, or “something to draw”
  • a bare noun or short phrase with no photo attached

The important distinction — do not confuse this with surprise-me:

  • surprise-me invents a random fictional CHARACTER (a person: hair, outfit, expression) and ignores whatever the user asked for. It is a portrait of a made-up stranger.
  • This skill draws the idea the user actually typed — usually an object, scene, or creature, not a person — and stays faithful to it.

So when the user names a thing, use this skill. When they explicitly want a surprise character or a random person, use surprise-me instead. For a doodle of a real person from a photo use doodle-avatar; for a real pet use pet-portrait.

What to draw

If the user described a subject at all, draw that subject and follow the description closely — the object, its setting, the little details they named. Keep it a single clear composition, not a collage.

If the user typed nothing, pick one interesting, drawable, concrete subject yourself — a plant, an animal, a food, a small object, a vehicle, a tiny scene. Deliberately do not default to a person or a portrait: the pages that use this skill are “doodle ideas”, “doodle prompt generator” and “random doodle generator”, where the visitor wants a fun thing to draw, not a picture of a stranger. Vary the subject run to run so repeat visitors get fresh ideas.

Style notes for this skill specifically:

  • naive marker-and-ink doodle, the same house look as the rest of the catalogue
  • bold clean outlines, simplified shapes, flat cheerful colour
  • warm off-white ground, one cohesive subject centred with breathing room
  • playful, hand-sketched charm — not a technical diagram, not clip-art

No photorealism, no 3D, no heavy realistic shading, no lettering.

How to run it

  1. Never ask for a photo — this skill needs none.
  2. Call generateDoodle with skill: "idea". Pass the user’s own words as description when they gave any; omit it to have the skill pick a subject.
  3. Never pass imageUrl for this skill — there is no source photo and the tool ignores it.
  4. Report the result briefly without repeating the URL.

Following up

Offer a twist on the same idea — a different palette, a new setting, or a companion object — or suggest another concrete subject to doodle next.

More skills

Browse all 40
Hand-drawn doodle avatar of a person with long wavy hair and round glasses
SAMPLE AVATAR
Nova hand-drawn character
Doodle AI

Make something
worth keeping.

Sign in to create, save, and revisit your doodles across devices.

Starter credits included

Drop a note

It lands on the roadmap board for the team to see.

Credits

Let's sort this out directly

Doodle AI is in its testing phase, so checkout isn't switched on yet. Message me and I'll add credits to your account by hand — usually within a few hours.

No payment details are collected anywhere in this flow.

Settings Your profile

Give your team a name

Anyone you invite will see this name.

Join a team

Paste the invite link a teammate sent you.

Invite your teammates

Choose a role, then copy the link and send it yourself.