hatch-pet

Generate Codex-compatible animated pet spritesheets from character art or text concepts.

Updated May 20, 2026
One-click install
npx skills add https://github.com/Keimoshi/Silverloft --skill hatch-pet-keimoshi
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: hatch-pet
Source: https://github.com/Keimoshi/Silverloft/tree/main/open-design-port/skills/hatch-pet
Command: npx skills add https://github.com/Keimoshi/Silverloft --skill hatch-pet-keimoshi

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires Pillow, and includes scripts (resource) and references (resource) components.

What problem does it solve? Creating a custom animated desktop pet requires producing a precise 8x9 sprite atlas with consistent character identity across nine animation states, which is tedious and error-prone to do by hand. This Skill automates the full pipeline from concept or reference images to a packaged, validated pet. ## Core Features & Use Cases - End-to-End Pet Generation: Delegates visual generation to the $imagegen system skill for a base pet plus nine animation row strips (idle, running, waving, jumping, failed, waiting, review), then assembles them deterministically. - Deterministic Atlas Pipeline: Bundled Python scripts extract 192x208 frames via chroma-key removal and connected-component detection, compose a 1536x1872 PNG/WebP atlas, and validate geometry. - QA and Packaging: Produces contact sheets, preview videos, validation reports, and a pet.json package staged under the Codex pets directory for import into the app. - Use Case: Ask for a pixel-art shiba pet with a pomegranate prop; the Skill generates the base art, all animation rows, runs QA, and outputs an importable spritesheet.webp and pet.json. ## Quick Start Hatch me a tiny pixel-art shiba pet, friendly and sitting upright with a small pomegranate prop, using the hatch-pet skill end-to-end.

Frequently Asked Questions about hatch-pet

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create a custom animated Codex pet?

Run the hatch-pet skill with a pet concept or reference images. It generates a base image and nine animation row strips via $imagegen, extracts frames, composes the atlas, and packages pet.json plus spritesheet.webp under the Codex pets directory.

What spritesheet format does the Codex pet atlas require?

The atlas must be a 1536x1872 PNG or WebP with an 8-column by 9-row grid of 192x208 cells. Used cells contain sprite frames and unused cells must be fully transparent, matching the row and frame counts in the animation-rows reference.

Can I hatch a pet without providing reference images?

Yes. For text-only requests, the skill generates a base pet image from your concept first, then uses that base as the canonical visual reference for grounding every animation row generation job.

Why did my pet run fail finalization with row QA errors?

Finalization blocks when inspect_frames reports errors such as wrong frame counts, empty cells, or chroma-key-adjacent pixels. Run queue_pet_repairs.py to reopen only the failed rows, regenerate them with $imagegen, and finalize again.

When is running-left mirrored instead of generated?

Running-left may be derived by mirroring running-right only after visual inspection confirms the pet is symmetric with no one-sided markings, props, or text. Otherwise it must be generated as a normal grounded $imagegen row job.