What does this skill do, and when should you use it?
Hatch Pet is a Codex workflow for creating animated pets and spritesheets from character concepts, reference images, or brand cues. It calls the installed $imagegen skill for base art and animation rows, then uses Python scripts for extraction, assembly, validation, and visual QA. A normal run produces a 9-state 8×9 atlas, contact sheet, motion previews, and a pet.json package. It depends on Codex-specific workflow and packaging conventions, so other clients require adaptation.
Reads user descriptions, reference images, or brand cues; performs lightweight web-based brand discovery when only a brand name is supplied; calls $imagegen for base art and animation rows; uses layout guides, Python scripts, and jq to process 192×208 frames, derive running-left when appropriate, extract frames, compose a 1536×1872 atlas, validate transparency and unused cells, and produce contact sheets, GIF previews, review JSON, spritesheet.webp, and pet.json.
- A designer needs to turn a character concept or reference image into a Codex pet.
- A team needs a mascot pet based on company or prospect cues without copying trademarked text or logos.
- A product or frontend team needs a non-pixel character converted into a nine-state animated atlas.
- An engineer needs deterministic checks for transparency, frame layout, motion semantics, and visual identity consistency.
How do you install this skill?
- Bare brand-name input triggers web search; confirm the permitted external-access scope and avoid sending sensitive brand or project context to external services.
- The workflow includes deletion of generated source files; confirm successful copying and retain backups because no unified rollback mechanism is documented.
- Verify $imagegen, Pillow, jq, network services, and local Codex paths in the target environment; mainland-China reachability is unproven.
- License, versioning, maintenance ownership, and update channels are unclear, so the skill should not be treated as a fully governed long-term production dependency.
- Shell / CLI
- Network access
- Local filesystem
Pythonjq$imagegen system skill
The skill is located at skills/.curated/hatch-pet/ in openai/skills. The README gives this installation command for Codex: $skill-installer hatch-pet; restart Codex afterward. The README also says the repository is deprecated and points to the OpenAI Plugins repository for current plugin examples. The supplied source does not include an individual LICENSE.txt.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/openai/skills.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/.curated/hatch-pet" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
In Codex, request an animated pet from a character description, reference image, or brand cue. The workflow prepares a run directory, generates base art and animation rows, performs deterministic atlas processing and validation, and packages pet.json with spritesheet.webp after visual QA. A bare brand name triggers brand discovery first; generation requires $imagegen and the referenced local scripts to be available.
What are this skill's strengths and limitations?
- Covers generation, repair, validation, visual QA, and packaging in one workflow.
- Supports pixel, plush, clay, sticker, flat-vector, 3d-toy, painterly, brand-inspired, and auto styles.
- Defines detailed constraints for transparency, motion semantics, identity consistency, frame layout, and unused cells.
- Separates visual generation from deterministic image processing and provides a clear output structure.
- Explicitly depends on Codex's $imagegen skill, pet packaging paths, and application state contract.
- A full run may require one base generation plus up to nine animation-generation jobs, increasing time and cost.
- Brand-only requests trigger web search, while the source does not document search permissions, cost, or platform coverage.
- No test suite, individual license text, or cross-platform validation evidence is supplied.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Hatch Pet Animation Workshop this page ✓ OpenAI · Official | 51 · Use with care | ★ 28k | 4mo ago | — |
| Brandkit Generation Skill | 50 · Use with care | ★ 96k | 11d ago | MIT |
| imagegen-frontend-mobile — Mobile App Visual Generation Skill | 63 · Recommended | ★ 96k | 11d ago | MIT |
| Slides Grab HTML Skill | 60 · Recommended | ★ 1.2k | 1mo ago | MIT |
| Banana Claude — Gemini Image Creative Director | 59 · Recommended | ★ 1.1k | 7d ago | MIT |
The README marks this repository as deprecated and recommends the OpenAI Plugins repository for current plugin examples. Teams building new skills or skill-only plugins should therefore evaluate that repository and its build guide; Hatch Pet remains the repository's specific workflow for animated pets.
How did FollowSkills review this skill?
The evidence constrains the image-generation path, requires grounded inputs for row generation, and requires explicit approval before mirror derivation; it also discloses web search for brand discovery and local cleanup of generated files. Points are deducted because web search, image generation, and file deletion still create external or persistent effects without a uniform confirmation model, least-privilege boundary, rollback plan, or sensitive-data policy; license metadata is also unknown.
The workflow specifies dependencies, retries, stop conditions, mirror prerequisites, and deterministic scripts, with failures generally tied to a state or prompt file. Points are deducted because key paths were not executed during this static review, no skill-local test suite or comprehensive CI evidence is shown, and availability of $imagegen, Pillow, jq, and the runtime environment is assumed rather than verified.
The evidence clearly describes users, input types, supported styles, nine animation states, brand and non-brand scenarios, and relatively precise triggering language. Points are deducted because non-fit boundaries are incomplete, Chinese-language support is not documented, and reliance on web search or image-generation services may reduce reachability for mainland-China users.
The skill includes layered documentation through SKILL.md, agent interface metadata, animation-row references, a contract, and a QA rubric, with substantial command examples. Points are deducted because the license is unknown and no explicit versioning policy, changelog, maintenance owner, or update path is shown; several install and dependency assumptions remain implicit in environment variables and preinstalled skills.
The evidence covers an end-to-end goal from concept or reference image through spritesheet, pet.json, QA media, and validation, supported by deterministic assembly and checking scripts. Points are deducted because no representative artifact or third-party execution evidence is supplied, so generation quality, animation readability, and direct package usability remain unverified.
The scripts, fixed atlas contract, frame counts, validation rules, and manifest structure are auditable and provide some traceability. Points are deducted because committed tests, real CI results, and independent corroboration are not shown, while the core image-generation behavior depends on unavailable runtime outputs.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →