agentleFS
alisadiq-aialisadiq-ai/favstash-ai-video-editing-skills/skills/3-styles-characters/cartoon-character/SKILL.md

cartoon-character

Skill75 starsChanged 40 days agoApache-2.0View on GitHub

Style for edit-video: a lip-synced 2D cartoon character that talks, blinks, frowns, lights up and swaps poses on the beat. Use Ali's cartoon or Stashy (FavStash's mascot) out of the box, or make the creator's own with any image AI. Works as the cursor in animated-explainer, as a corner presenter on any HyperFrames reel, or big at the bottom of a cartoon explainer.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Write a schedule: poses on beats, brow windows, bounces on key words (example).

    line 51
  2. Distinctive:

    Bounce on four to six key words,: not on every line.

    line 64

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Cartoon character

A style on top of edit-video. It is a cut-out rig, not generated video: a handful of full poses plus small face patches (mouth shapes, closed eyes, brows), switched by one GSAP timeline from the voice. Every frame is seekable, renders are free, and the character looks the same in every reel.

Characters

CharacterFolderPoses
Ali, the creator of this pack, as a cartooncharacters/ali/rigneutral, point, thumbs, wave, shrug
Stashy, FavStash's mascotcharacters/stashy/rigneutral, point, thumbs, wave, shrug, explain
The creator's owntheir own folderwhatever they make

Both samples have mouth shapes m-ah, m-eh, m-oo, resting mouths m-grin and m-frown, e-closed for blinks, and u-worried / u-excited brows. They are for trying the skills; read the terms before using one in a published video. For a creator's channel, make their own character.

Make your own character with AI

Any image model that can edit an image while keeping its subject works: GPT Image, Nano Banana, Seedream, Flux Kontext and others. The full recipe and prompts are in make a character. In short:

  1. One base image of the character, front-facing from the chest up, on a flat green background. Use the creator's photos as references if it's them.
  2. Edits of that base: four or five poses, then three mouth shapes, closed eyes and two expressions, each changing only that part.
  3. scripts/build_rig.py keys out the green, lines every edit up with the base, cuts the face patches and writes the rig folder with an alignment report.

A rig is 1024×1024 PNGs plus patches.json. New poses and patches need no code: the presenter picks up every pose-*.png and every entry in patches.json.

Use it

  • In animated-explainer: set CHARACTER = 'ali', 'stashy' or a rig folder in the spec. The builder runs the steps below and puts the character in the cursor badge, or on the rail head with STYLE = 'rail'.
  • In any HyperFrames composition (a corner presenter on a talking-head edit, or big at the bottom of a cartoon explainer):
    1. python3 scripts/lipsync.py --audio voice.wav --words words.json --out mouth.json turns the voice (and its word times, if you have them) into mouth shapes.
    2. Write a schedule: poses on beats, brow windows, bounces on key words (example).
    3. python3 scripts/presenter.py --rig <rig> --rig-url assets/presenter --schedule schedule.json --mouth mouth.json --out-html presenter.html --out-pres pres.json [--left --top --size]
    4. Copy the rig to assets/presenter/, load assets/presenter.css and assets/presenter.js, paste presenter.html into the composition root and call addPresenter(tl, PRES) with the contents of pres.json, after the scene tweens.

Acting

  • Pose on meaning: shrug on the problem, point while explaining or driving the UI, thumbs on a win, wave on the call to action. Hold a pose at least a second.
  • Brows tell the arc: worried before the turn, excited after it.
  • Bounce on four to six key words, not on every line.
  • Mouth only on its own words. With two voices, pass only this character's words to lipsync.py, so it stays quiet while the other one talks.
  • Small and in the scene. A badge of about 200 px or a corner presenter of about 650 px. It never covers text being read.
  • The voice matches the character. The creator's own voice, or a clone of it with their OK. See animated-explainer's voice options for free local voices.

Pitfalls already paid for

  • Two tl.set calls on one element at time 0 get reverted when the renderer seeks back to 0. The opening pose and mouth use gsap.set, which presenter.js does.
  • An image model that redraws slightly between edits leaves patches that don't sit on the face. Run build_rig.py --register and read the alignment report before trusting a rig.
  • Prompting "keep everything pixel-identical" stops some models from changing the part you asked for. Name the one thing to change and say everything else stays.
  • Colour drifts across many edits (hair turning pink after a dozen). Compare every edit with the base and correct it before building the rig.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.