Turn the sprite you just made into a gallery pull request against with-pebbly/aseprite-ai-artist — exported files, an honest generation.yaml (prompts verbatim, model, plugin version, validate…

MITAuto-check passedGame Development

Install Submit

skills CLI
$ npx skills add with-pebbly/aseprite-ai-artist --skill submit -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install with-pebbly/aseprite-ai-artist submit --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/with-pebbly/aseprite-ai-artist.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/submit .claude/skills/submit && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
submit
GitHub stars
172
Token cost
~3.1k tokens
SKILL.md length
1,583 words
Files
1
Skills in repo
14
Repo updated
First seen
Licence
MIT

At a glance

Turn the sprite you just made into a gallery pull request against with-pebbly/aseprite-ai-artist — exported files, an honest generation.yaml (prompts verbatim, model, plugin version, validate…

  • Works in 7 steps: Preconditions → Gather metadata → Benchmark mode → …
  • The user wants to share
  • SKILL.md covers 1. Preconditions, 2. Gather metadata, 3. Benchmark mode and 4. Export, plus 5 more sections
  • Calls gh, git and pnpm; reaches cli.github.com

What it does

Submit is an agent skill from with-pebbly/aseprite-ai-artist. Turn the sprite you just made into a gallery pull request against with-pebbly/aseprite-ai-artist — exported files, an honest generation.yaml (prompts verbatim, model, plugin version, validate results, optional benchmark scoring), checked with pnpm gallery:check, opened with the gh CLI. Use when the user wants to share, publish or submit their work to the gallery or the benchmark.

Its SKILL.md is about 3.1k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Game Development, covering Sprites and pixel art and Pull requests. It works with pnpm and Model Context Protocol. The repository describes itself as: Your coding agent paints pixel art in the Aseprite window you already have open. MCP server + skills for Claude Code, omp, Codex, Gemini CLI and Cursor — draws, looks at its…. The licence is MIT.

When your agent uses it

  • The user wants to share
  • Submit their work to the gallery

Example prompts

  • “/submit”

Requirements

  • Node.js

Workflow steps

7 steps, taken from the step headings in SKILL.md.

  1. Preconditions
  2. Gather metadata
  3. Benchmark mode
  4. Export
  5. Write generation.yaml
  6. Validate in a clone
  7. Submit

What it can do on your machine

Read from SKILL.md and the folder at commit f51fe69. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • gh
    • git
    • pnpm
    • npx

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    Hosts in commands or code, which the agent is likely to contact:

    • cli.github.com

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Submit loads about 3.1k tokens when it runs. Until then it costs about 98 tokens; SKILL.md has 1,583 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~98
When it runs · the whole SKILL.md, loaded when a task matches
~3.1k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from with-pebbly/aseprite-ai-artist at commit f51fe69, republished under its MIT licence (© with-pebbly). 1,583 words, ~3,085 tokens.

Download SKILL.mdSave it as .claude/skills/submit/SKILL.md (or your agent's skills folder).
name
submit
description
Turn the sprite you just made into a gallery pull request against with-pebbly/aseprite-ai-artist — exported files, an honest generation.yaml (prompts verbatim, model, plugin version, validate results, optional benchmark scoring), checked with `pnpm gallery:check`, opened with the gh CLI. Use when the user wants to share, publish or submit their work to the gallery or the benchmark.
title
Submit a finished sprite to the community gallery

The gallery keeps how a sprite was made next to the sprite: the prompts as sent, the model, the plugin version, what validate said. A submission that tidies any of that up is worse than none — the benchmark compares models on it. Record what happened, not what you wish had.

The contract is gallery/src/schema.ts in the repository; pnpm gallery:check enforces it. This workflow produces files that pass it.

1. Preconditions

Check all three before touching anything. If one fails, stop and tell the user exactly what to do; never route around it.

  • preflight is ready. Not ready → stop. Never export or edit files on disk instead of the live document.
  • The work is finished and reviewed. Run aseprite:review if it has not been. Submitting art that fails review puts it in public.
  • gh is installed and authenticated. Run gh auth status. Missing → tell the user to install it (https://cli.github.com) and run gh auth login. There is no other route to a pull request.

Then ask for consent, once, before anything leaves the machine: "I will fork with-pebbly/aseprite-ai-artist on your GitHub account, push a branch, and open a public pull request with your sprite and the prompts used. OK?" Forking, pushing and opening a PR act as the user, in public. No yes, no fork.

2. Gather metadata

Take what the session already knows. Ask the user only for the rest, in one message.

FieldSource
title, descriptionAsk. Offer a title drawn from the work.
author.name, author.githubgh api user --jq '.name,.login' as the default; let the user override. Handle without @.
harnessYou know which you run in: claude-code, omp, codex, gemini-cli, cursor…
modelsThe model id exactly as the harness names it (claude-opus-4-1, gpt-5). Not a nickname. If different models ran different steps, list each once and name the model on every step.
pluginpreflight returns extensionVersion — the plugin build attached to Aseprite. Use it. If its directive says the extension and the server differ, the run mixed two builds and has no single version: stop and tell the user to align them (install-extension, restart Aseprite) before submitting. If extensionVersion is null, run npx @pebbly/aseprite-ai-artist --version. It must be a released version listed in the repo's CHANGELOG.md; a dev or unreleased build cannot be submitted — say so.
steps[].textEvery user prompt that shaped the work, verbatim and in order, copied from the conversation. Never paraphrase or merge. A prompt sent in another language may be rendered in English here only if the verbatim text goes into steps[].original — never for a benchmark run.
steps[].interventionsEach clarifying question you asked and the answer the user gave, plus any steering beyond the prompt. Empty if none.
referencesWhere the design came from — required, never guessed. source: none when the work was drawn from words alone (the user chose "continue without" in aseprite:concept, or it was never offered). generated when an image model made a concept sheet or storyboard: ask which model if you do not know, list it in imageModels, the kinds in kinds, and put the image prompt you wrote in prompt. supplied when the user handed over their own art. Copy the reference image into the folder as a reference file when you have it.
dateToday, yyyy-mm-dd.
tagsA few lowercase words: character, animation, 32x32.

For each step run validate on the state the step produced (or on the final document if steps were not saved separately) and record errors, warnings, score and passed as the tool reported them. Do not round a failing report into a pass.

3. Benchmark mode

Only if the user says the run followed a benchmark prompt and the user is a maintainer. Benchmark runs are the project's own measurements; the checker and CI refuse a benchmark: block from anyone else. Check first: take the login from gh api user --jq .login (never from what the user typed or the YAML), and look for it, case-insensitively, in https://raw.githubusercontent.com/with-pebbly/aseprite-ai-artist/main/gallery/MAINTAINERS. Not listed, or either call fails → not a maintainer: say benchmarks are run by the maintainers, skip this section and submit a gallery generation. A maintainer sets author.github to that same login.

  1. Fetch its definition: the local clone's gallery/prompts/<id>/prompt.yaml, or https://raw.githubusercontent.com/with-pebbly/aseprite-ai-artist/main/gallery/prompts/<id>/prompt.yaml.
  2. Confirm the run followed the fixed setup (canvas, palette, the rules list) and that every step's text equals the prompt's step text word for word — only surrounding whitespace and line endings are ignored. Any difference — a reworded step, extra steering, a reference image — means this is not a benchmark run. Submit it as a free generation (omit benchmark) and tell the user why.
  3. Judge every criterion in the prompt, none skipped. Get the evidence first: look op preview for the read, a filmstrip or onion-skinned view for motion, validate (with expect where the criterion names a contract), tag op="list", frame op="list". Record pass and a one-line note naming what you saw. Never mark a pass on a criterion whose evidence you did not look at in this session; look now, or mark it failed and say why.
  4. Write benchmark: {prompt, revision, results} with the prompt's revision and one result per criterion id.
Show full SKILL.md (724 more words)Show less

4. Export

Stage everything in one flat folder named <yyyy-mm-dd>-<slug>, slug lowercase with single hyphens. Use the plugin's own export; never write these by hand.

  1. Save the live document, then keep a copy: sprite_manage op="save", then export op="aseprite" path="<folder>/<slug>.aseprite". Role source. The gallery keeps the editable original; never copy or edit a .aseprite file from disk.
  2. Cover. export op="png" scale=N frame=1 path="<folder>/<slug>.png" with the smallest integer N that makes the long side at least 256 px (32×32 → 8), capped at the tool's maximum of 16 — a smaller sprite simply gets a smaller cover. For an animation, the GIF can be the cover instead. Role cover; exactly one file.
  3. Animation (more than one frame): export op="gif" scale=N path="<folder>/<slug>.gif", same scale. Role animation, or cover if it is the cover.
  4. Filmstrip (animated): export op="spritesheet" sheetType="horizontal" scale=N includeJson=false path="<folder>/<slug>-strip.png", same scale. Role filmstrip.
  5. Per step. When steps produced separate documents or intermediate saves, export the source (and a still) for each and set step on those files.

Files stay flat — no subfolders — and only .aseprite, .png, .gif, .webp or .json. Each file must be under 95 MiB (GitHub hard-blocks 100); anything over 50 MiB draws a warning, so keep covers and GIFs small.

Look at the cover and the GIF's first frame before continuing: what goes into the public gallery is what you exported, not what you remember drawing.

5. Write generation.yaml

In the staging folder, exactly per the schema:

yaml
title: Cherry tree in bloom
description: A 64×64 cherry tree drawn to the tree-growth brief.
date: 2026-09-29
author: { name: Ada Lovelace, github: ada }
plugin: 0.3.2
harness: claude-code
models: [claude-opus-4-1]
tags: [nature, animation]
references:
  source: generated
  imageModels: [gpt-image-2]
  kinds: [concept-sheet]
steps:
  # the step text exactly as sent — for a benchmark run, verbatim from
  # prompt.yaml (cut short here)
  - text: |-
      The document is already open: 64×64 pixels, RGB, with the PICO-8 palette loaded (16 colours). …
    interventions: []
files:
  - { path: tree.png, role: cover, label: The tree, step: 1 }
  - { path: tree.aseprite, role: source, label: Source document }
validate:
  - { step: 1, passed: true, score: 92, errors: 0, warnings: 1 }
benchmark:            # only for a confirmed benchmark run
  prompt: tree-growth
  revision: 1
  results:
    - { criterion: tree-validate, pass: true, note: "validate: 0 errors, 1 warning, named." }

Rules that fail the check when missed:

  • Exactly one cover (png/gif/webp) and at least one .aseprite with role source; every file listed once, nothing unlisted in the folder.
  • With more than one model, every step names its model, and it appears in models.
  • step values and validate[].step are 1-based and must exist.
  • benchmark.results names criterion ids that exist in the prompt.

Two optional blocks:

  • metrics — per step, from the session you can see: minutes, toolCalls, outputTokens, costUsd. Fill only what the harness actually reports; never estimate.
  • ratings — leave them out. You took part in this run, and the checker refuses model:<id> for any of the run's own models. Judges add ratings later, blind, against gallery/RUBRIC.md.

Show the user the final YAML and the file list, and wait for a go-ahead. They are about to publish it under their name.

6. Validate in a clone

  1. Use a clone of the user's fork if one exists; otherwise fork and clone (gh repo fork with-pebbly/aseprite-ai-artist --clone) — consent from step 1 covers this. Make sure the checkout is up to date with upstream main.
  2. Copy the staging folder to gallery/generations/<folder>/.
  3. pnpm install if node_modules is missing, then pnpm gallery:check.
  4. Fix every error and re-run until clean. Fix by correcting the YAML or re-exporting, never by editing the recorded truth (prompts, results) to satisfy the checker. Warnings: read them, mention them to the user.

Never submit with errors.

7. Submit

git switch -c gallery/<folder>
git add gallery/generations/<folder>
git commit -m "gallery: <title>"
git push -u origin gallery/<folder>
gh pr create --repo with-pebbly/aseprite-ai-artist --head <login>:gallery/<folder> --title "gallery: <title>" --body-file <filled body>

Build the body from .github/PULL_REQUEST_TEMPLATE/generation.md in the clone: keep its sections and checklist, fill them honestly — what the sprite is, harness, model, plugin version, whether it is a benchmark run and its score — and pass that file as --body-file. Before pushing, git status must show only the one folder as added; anything else in the diff means stop and clean it up. A PR adds its own folder and nothing more.

Report the PR URL. Tell the user CI will re-run gallery:check and that a maintainer reviews it.

Failure modes

SymptomDo
preflight not ready / not_connectedStop. Relay the tool's directive. Do not touch files on disk.
gh missing or not logged inStop. Give the install and gh auth login steps.
User declines consentStop; leave the staging folder for them and say where it is.
pnpm gallery:check errorsFix the YAML or files, re-run. Never submit red.
A file exceeds the size limitLower the export scale, drop the GIF to a shorter loop, or use a PNG cover; never strip the .aseprite.
Prompt text differs from the benchmark'sNot a benchmark run. Drop benchmark, submit as a free generation, say why.
Plugin version not in CHANGELOG.mdIt is an unreleased build. Tell the user; they can submit after installing a released version.
Push or PR creation failsReport the gh error as printed; do not retry with other credentials or routes.

aseprite:review before submitting, aseprite:export for export options, rules://07-review-checklist.

© with-pebbly, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in skills/submit of with-pebbly/aseprite-ai-artist.

Open the folder on GitHubat commit f51fe69

Compare with similar skills

Submit next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Submit compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Submit this skillwith-pebbly/aseprite-ai-artist172—~3.1kAutomated safety check: PassMIT
Game Art PipelinePersonalJarvis/PersonalJarvis141—~566Automated safety check: PassApache-2.0
Sprite Genaldegad/sprite-gen2.6k—~4.5kAutomated safety check: PassApache-2.0
Fal Assetsrehan-remade/universal-modder5.3k—~2kAutomated safety check: NotesMIT
Game Assetsglifxyz/glif-mcp-server212—~831Automated safety check: PassMIT
Scenario Game Assetsscenario-labs/skills913—~4.5kAutomated safety check: PassMIT

Similar skills

  • Game Art Pipeline

    PersonalJarvis/PersonalJarvis

    Develop or redesign Personal Jarvis world art, buildings, characters and 3D assets through an authored Blender reference scene, runtime review and controlled asset rollout.

    141 GitHub stars~566 tokensUpdated today
    Game DevelopmentAuto-check passed
  • Sprite Gen

    aldegad/sprite-gen

    Generates images and game sprites through GPT or Grok with guided provider choices, separate saved defaults, automatic cleanup and optional curation.

    2.6k GitHub stars~4.5k tokensUpdated today
    Game DevelopmentAuto-check passed
  • Fal Assets

    rehan-remade/universal-modder

    Generate game assets with fal (fal.ai) through the fal MCP server, the um fal CLI (REST) or fal api.

    5.3k GitHub stars~2k tokensUpdated today
    Game DevelopmentAuto-check: notes
  • Game Assets

    glifxyz/glif-mcp-server

    Make art and media for a game you are building with Glif: character reference sheets, consistent characters, pixel-art sprite sheets with animation cycles, seamless and PBR textures, 3D models, item…

    212 GitHub stars~831 tokensUpdated today
    Game DevelopmentAuto-check passed
  • Scenario Game Assets

    scenario-labs/skills

    A skill your agent uses when creating game art through Scenario MCP: sprites, sprite sheets, icons, props, loot, tilesets and autotile terrain transitions, seamless tiles, isometric buildings…

    913 GitHub stars~4.5k tokensUpdated yesterday
    Game DevelopmentAuto-check passed
  • Pull Requests

    pnpm/pnpm

    Take a change through a pull request in the pnpm repository — opening it, then staying with it after every push until CI is green and the review round is quiet.

    37k GitHub stars~2.2k tokensUpdated today
    DevelopmentAuto-check passed

More from with-pebbly/aseprite-ai-artist

All 14 skills in this repo
  • Animate

    with-pebbly/aseprite-ai-artist

    Build an idle, walk, run or attack cycle from key poses, set timing that reads as motion rather than a metronome, tag it, and review it as a filmstrip.

    172 GitHub stars~1.4k tokensUpdated today
    Auto-check passed
  • Draw

    with-pebbly/aseprite-ai-artist

    Draw a sprite from silhouette to finished pixels, in the order that catches mistakes while they are still cheap — block in, check the silhouette, shade, outline, verify.

    172 GitHub stars~1.5k tokensUpdated today
    Auto-check passed
  • Export

    with-pebbly/aseprite-ai-artist

    Produce the files an engine actually consumes — spritesheets with atlases, GIFs, scaled PNGs — with the tags and layout the target needs.

    172 GitHub stars~628 tokensUpdated today
    Auto-check passed
  • Fix

    with-pebbly/aseprite-ai-artist

    Change a sprite that already exists — the user's own work or your earlier output — without destroying what is already right.

    172 GitHub stars~1.1k tokensUpdated today
    Auto-check passed
  • New

    with-pebbly/aseprite-ai-artist

    Create a new Aseprite document with the right canvas size, colour mode, palette and layer structure, so later work does not have to fight the setup.

    172 GitHub stars~555 tokensUpdated today
    Auto-check passed
  • Palette

    with-pebbly/aseprite-ai-artist

    Pick a palette that fits the request, build hue-shifted ramps, or clean up a sprite whose colours have sprawled.

    172 GitHub stars~795 tokensUpdated today
    Auto-check passed

Questions about Submit

What does Submit do?

Turn the sprite you just made into a gallery pull request against with-pebbly/aseprite-ai-artist — exported files, an honest generation.yaml (prompts verbatim, model, plugin version, validate…. Submit is an agent skill from with-pebbly/aseprite-ai-artist.yaml (prompts verbatim, model, plugin version, validate results, optional benchmark scoring), checked with pnpm gallery:check, opened with the gh CLI.

When should I use Submit?

Submit fits situations like: the user wants to share; submit their work to the gallery.

How do I install Submit in Claude Code?

Run `npx skills add with-pebbly/aseprite-ai-artist --skill submit -a claude-code`. Or copy the skill folder (skills/submit in with-pebbly/aseprite-ai-artist) into .claude/skills/submit in your project. Claude Code loads it when a task matches its description.

How do I install Submit in Codex?

Run `npx skills add with-pebbly/aseprite-ai-artist --skill submit -a codex`. Or copy the skill folder (skills/submit in with-pebbly/aseprite-ai-artist) into .agents/skills/submit in your project. Codex loads it when a task matches its description.

Can I use Submit in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add with-pebbly/aseprite-ai-artist --skill submit -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/submit, .gemini/skills/submit, .github/skills/submit and .opencode/skills/submit in your project.

What does Submit need to run?

Going by SKILL.md and its folder, Submit needs the command-line tools its instructions call (gh, git, pnpm and npx). Our summary lists: Node.js.

Does Submit access the network?

SKILL.md names 1 domain. In commands or code: cli.github.com; the agent is likely to contact it when it follows the instructions. This is read from the text; nothing was executed.

Is Submit safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Submit use?

Submit is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Submit use?

About 3.1k tokens (SKILL.md is roughly 12k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Submit?

Skills that share tags, products or a category with Submit: Game Art Pipeline (PersonalJarvis/PersonalJarvis, 141 stars), Sprite Gen (aldegad/sprite-gen, 2.6k stars), Fal Assets (rehan-remade/universal-modder, 5.3k stars) and Game Assets (glifxyz/glif-mcp-server, 212 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Submit?

with-pebbly (a GitHub organization) maintains it in with-pebbly/aseprite-ai-artist, which has 172 GitHub stars. The repository holds 14 skills in this directory. The repository was last updated on October 8, 2026.

Source: with-pebbly/aseprite-ai-artist on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.