logostickers

Give your logo
a little life.

An open-source agent skill that turns a real logo into a character with something to do. Each one here started as a flat logo.

Qunxue: four charcoal paper researchers examining a daisy

Qunxueshape → cast

MultiPush: a blue octopus typing, folding paper and drinking coffee at once

MultiPushmeaning → ability

Windup: a mechanical bird taking off with its winding key

Windupmark → motion

Drag a corner, or just click.

npx skills add huyanxius/logo-to-stickers

All fifteen stickersRead the skill

Three marks, fifteen little stories

These came out of one night of making stickers for a portfolio. The early attempts were stiff. Things started working when the brand itself supplied the cast, the ability, and the small things that could go wrong.

Original prompts Full-size PNGs

What the skill pays attention to

Give a brand a small cast, a useful ability, and something slightly surprising to do. Each sticker gets one sentence of plot:

who does what→a small unexpected consequence→a readable reaction

  1. 01

    Read the real mark

    Keep the shapes, component count, or colour relationship that makes the brand recognisable.

  2. 02

    Choose a translation

    Let logo parts become a cast, or turn a product behaviour into a character’s ability.

  3. 03

    Give it a situation

    An action, a small consequence, and a reaction that still reads at thumbnail size.

  4. 04

    Keep one identity

    Reuse the first good sticker as the character reference while the action changes.

  5. 05

    Finish it as paper

    Matte printed colour, a narrow white die-cut edge, and real transparency outside it.

How the three brands were translated

Make your own set

  1. 1Add the skill

    npx skills add huyanxius/logo-to-stickers

    Works with any agent the Skills CLI supports. Add --list to look inside first.

  2. 2Bring your logo

    Attach the actual mark and ask:

    Use $logo-to-stickers to make five playful paper stickers from this logo. Keep the brand recognisable, give each image a little story, and save five separate transparent PNGs.

  3. 3Meet the cast

    The agent makes one scene first, checks it, then uses it as the identity anchor for the rest of the set.

    Full guide

Tested with Codex’s built-in image generator. Your agent needs an image tool that accepts reference images and produces transparent output; the skill brings the workflow, not an image model.