Back to Skills

hatch-pet

Create, repair, validate, preview, and package Codex-compatible animated pet spritesheets from character art, screenshots, generated images, or visual references. Use when a user wants to hatch a Codex pet, create a custom animated pet, or build a built-in pet asset with an 8x9 atlas, transparent unused cells, row-by-row animation prompts, QA contact sheets, preview videos, and pet.json packaging. This skill composes the installed $imagegen system skill for visual generation and uses bundled scr

76,147stars8,682forksUpdated 7/8/2026

Security Assessment

Safe(100/100)
Security Score100/100

About hatch-pet

Hatch Pet creates, repairs, validates, previews, and packages Codex-compatible animated pet spritesheets from a concept, reference images, screenshots, or generated art. The end product is an 8x9 sprite atlas (192x208 per cell) with transparent unused cells, a pet.json manifest, QA contact sheets, and preview media, packaged so the sprite can be imported into the Codex floating pet companion via Settings, Pets, Import Codex sprite, which auto-detects the atlas and lets the user pick an animation row to play (idle, running-right, waving, and so on).

It owns the pet-specific work (prompt planning, animation rows, frame extraction, atlas geometry, QA, previews, and packaging) while delegating all normal visual generation to the installed $imagegen system skill rather than calling any image API directly. A normal run expects up to ten generation jobs: one base pet plus nine row strips, with running-left optionally derived by mirroring an already-approved running-right. A firm boundary forbids fabricating pet visuals with local Python/Pillow, SVG, canvas, or HTML/CSS; the bundled scripts are only for deterministic processing of already-generated outputs, recorded through record_imagegen_result.py (or generate_pet_images.py as a documented secondary fallback).

Use it when a user wants to hatch a Codex pet or build a custom animated pet asset. User inputs (pet name, description, reference images) are optional and inferred when omitted. The default art style deliberately matches Codex's built-in digital pets: compact chibi pixel-art-adjacent mascots with chunky silhouettes, thick outlines, limited palettes, and flat cel shading, simplifying more detailed references into that house style. Detailed transparency-and-effects rules keep every pixel either part of the sprite or cleanly chroma-keyable, allowing only small attached, hard-edged effects and disallowing motion lines, detached sparkles, shadows, glows, text, and backgrounds. This is the unmodified upstream Codex hatch-pet skill vendored for Open Design agents.

FAQ

What files does a completed pet include?

The primary output is final/spritesheet.png, plus final/spritesheet.webp, a pet.json manifest, and a qa/contact-sheet.png. The webp can be imported into the Codex pet companion.

How are the sprites actually generated?

Visual generation is delegated to the installed $imagegen system skill, never the Image API directly. The skill's own bundled scripts only do deterministic work like frame extraction, validation, atlas composition, and packaging.

What are the atlas dimensions?

An 8x9 atlas with 192x208 cells, with unused cells left transparent. The import flow auto-detects the 8x9 / 192x208 atlas.

What art style does it produce?

The Codex digital-pet house style: small pixel-art-adjacent chibi mascots with chunky silhouettes, thick 1-2px outlines, limited palettes, and flat cel shading, simplifying more detailed references into that style.

Can I skip providing a pet name or reference images?

Yes. All user-facing inputs are optional. The skill infers a name and description from the concept or filenames, and if no references are given it generates a base pet from text and uses that as the canonical reference for each row.

All Files

23 files
agents/openai.yaml0.5 KB
View
references/codex-pet-contract.md0.8 KB
View
scripts/extract_strip_frames.py10.4 KB
View
scripts/make_contact_sheet.py2.9 KB
View
scripts/queue_pet_repairs.py6.6 KB
View
scripts/render_animation_videos.sh0.1 KB
View
SKILL.md23.3 KB
View
README.md2.6 KB
View
scripts/compose_atlas.py5.1 KB
View
scripts/generate_pet_images.py10.2 KB
View
scripts/pet_job_status.py4.0 KB
View
scripts/render_animation_videos.py4.1 KB
View
scripts/test_generate_pet_images.py5.0 KB
View
LICENSE.txt10.5 KB
View
references/qa-rubric.md3.2 KB
View
scripts/finalize_pet_run.py13.7 KB
View
scripts/package_custom_pet.py3.4 KB
View
scripts/record_imagegen_result.py7.7 KB
View
references/animation-rows.md1.4 KB
View
scripts/derive_running_left_from_running_right.py4.8 KB
View
scripts/inspect_frames.py8.3 KB
View
scripts/prepare_pet_run.py26.8 KB
View
scripts/validate_atlas.py4.7 KB
View

Install hatch-pet

Download and extract the skill files to your .claude/skills/ directory.

Quick Setup:

  1. Copy the skill folder to .claude/skills/
  2. Claude will automatically detect and use the skill