Search the OpenPets community registry for Codex / Claude Code / OpenCode / Pi Code pets matching user criteria, returning ranked pet metadata and universal one-click install links for the OpenPets macOS app.
Works with
AI-first code editor with Composer
Before installing skills in Cursor, ensure your development environment meets these requirements:
node --versionsearch-codex-petsExecute the skills CLI command in your project's root directory to begin installation:
Fetches search-codex-pets from openpets.sh/search-codex-pets-f0oals and configures it for Cursor.
The CLI shows a list of agents. Use arrow keys and space to select Cursor:
Confirm successful installation by checking the skill directory location:
Restart Cursor to activate search-codex-pets. Access via /search-codex-pets in your agent's command palette.
We perform automated surface-level scans (Gen AI Scanner, Socket, Snyk) during installation. These checks detect common vulnerabilities but do not guarantee complete security. Always review skill source code and verify the publisher's reputation before production use.
Skills execute code in your environment. Always review source, verify the publisher, and test in isolation before production.
Submit your Claude Code skill and start earning
Automate repetitive workflows and reduce manual effort
Example
Generate reports, summarize documents, draft communications
Save 3-5 hours per week on routine tasks
Learn new skills, understand complex topics, get expert guidance
Example
Explain concepts, provide examples, suggest learning resources
Accelerate learning and skill development by 2x
Enhance output quality through reviews, suggestions, and refinements
Example
Review drafts, suggest improvements, catch errors
Improve work quality by 30-40% with less effort
0
total installs
0
this week
0
upvotes
Run in your terminal
0
installs
0
this week
—
stars
| name | search-codex-pets |
| title | Search OpenPets Codex Pet Registry |
| description | >- Search the OpenPets community registry for Codex / Claude Code / OpenCode / Pi Code pets matching user criteria, returning ranked pet metadata and universal one-click install links for the OpenPets macOS app. |
| website | openpets.sh |
| category | developer-tools |
| tags | - codex - openpets - registry - search - ai-tools - macos |
| source | 'browserbase: agent-runtime 2026-05-19' |
| updated | '2026-05-19' |
| recommended_method | api |
| alternative_methods | - method: browser rationale: >- Drive the public gallery UI (form#gallery-form on openpets.sh) as a fallback if /api/pets becomes unreachable. Strictly slower due to IntersectionObserver-driven lazy loading; the JSON API returns the same records in one round trip. |
| verified | true |
| proxies | true |
Anonymously search the OpenPets community registry — the largest catalog of pixel-art "pets" (desktop companions) built for Codex, Claude Code, OpenCode, and Pi Code — and return ranked pet metadata plus a universal one-click install link suitable for handoff to the OpenPets macOS app. Read-only. No authentication required. Returns JSON.
animal, creature, object, or person.The OpenPets website exposes a public, paginated REST endpoint at /api/pets that powers the on-page gallery. Always prefer this over scripted browsing — no auth, no rate limits observed, returns the full record set including install-link material in one round trip.
Issue a single GET to the search endpoint:
GET https://openpets.sh/api/pets?q={query}&kind={kind}&sort={sort}&page={n}&pageSize={n}
Accept: application/json
Map user criteria to query params:
| Param | Type | Required | Notes |
|---|---|---|---|
q | string | no | Full-text search over displayName, description, and tags. URL-encode it. |
kind | enum | no | One of animal, creature, object, person. Any other value returns zero results (total: 0). |
sort | enum | no | One of new (default), popular, liked. Invalid values silently fall back to new. |
page | int | no | 1-indexed. Default 1. |
pageSize | int | no | Default ~24. pageSize=100 works; upper bound undocumented. |
Parse the JSON response. Top-level shape:
{ "page": 1, "pageSize": 24, "total": 239, "totalPages": 10,
"generatedAt": "2026-05-18T11:13:49.412Z", "pets": [ ... ] }
For each pet in pets[], extract:
id — slug used in URLs (e.g. qwq, apupepe, aobing)displayName, description, kind, ownerName, tags[]viewCount, downloadCount, likeCountpreviewUrl (relative /api/pets/{id}/preview) — animated WebP previewspritesheetUrl (relative /api/pets/{id}/spritesheet) — full sprite atlasshareImageUrl — 1200×630 OpenGraph cardConstruct the install link for each pet:
https://openpets.sh/install/{pet.id}
This URL is public and anonymous. Hitting it returns a 302 with:
Location: openpets://install?url=<signed-download-url-with-ticket>&id={pet.id}
The signed ticket is a server-issued JWT-style token with a ~24h expiry (exp, nonce, id payload). The OpenPets macOS app intercepts the openpets:// protocol and pulls the bundle. Always hand out the https://openpets.sh/install/{id} URL — never try to construct the openpets:// deep-link yourself (the ticket is server-signed).
Optional detail enrichment: GET https://openpets.sh/api/pets/{id} returns { "pet": { ... } } with the same record shape if you need a single record without searching.
Return the ranked list with id, displayName, description, kind, ownerName, tags, stats, previewUrl, and installUrl.
If /api/pets ever becomes unreachable, drive the gallery UI directly. This is strictly slower (lazy-loaded sprites, IntersectionObserver pagination) but the contract is stable:
https://openpets.sh. Optionally pre-seed query state via URL params (?q=cat&kind=animal&sort=popular) — the gallery script reads them on load.input#q with the search term, set select#sort and select#kind.button[type="submit"] inside form#gallery-form.#gallery-results to populate. Each card is <article class="pet-card"> with a link <a class="pet-card-preview-link" href="/pets/{id}">.href attribute. Construct install URLs as https://openpets.sh/install/{id}.#gallery-sentinel into view or click #load-more-pets.No proxies or stealth required for either path — the site is behind Cloudflare but the API and gallery are CORS-open and bot-tolerant.
Two response shapes from /api/pets depending on sort order. With sort=new (the default), the response is served from a registry mirror and includes lighter, mirror-flavored fields: installTicketUrl, downloadUrl, validationReport, source.apiBase, mirroredAt, registryNumber. With sort=popular or sort=liked, the response is served live from the primary database and includes richer fields: spritesheetPath, ownerId, ownerHandle, likedByMe, reactionCounts, myReactions, ownerShadowbanned. Both share the core set (id, displayName, description, kind, ownerName, tags, uploadedAt, viewCount, downloadCount, likeCount). Don't depend on the extended fields being present.
Invalid kind ⇒ empty results; invalid sort ⇒ silent fallback. kind=foo returns total: 0; sort=foo returns the full registry sorted by new. Validate kind against the enum before sending or you'll mislead the user with "0 matches".
installTicketUrl and downloadUrl in the response require auth. POST /api/pets/{id}/install-ticket returns 404 anonymously, and GET /api/pets/{id}/download returns 401 {"error":"download ticket required"}. Ignore those fields for anonymous flows. The public /install/{id} redirect is the supported anonymous path; it generates the signed ticket server-side and embeds it in the openpets:// Location header.
/install/{id} requires the OpenPets macOS app to actually resolve. The openpets:// URI scheme is registered by the app from the alterhq/openpets release. Users on Linux/Windows, or macOS without the app, will see "no app to handle this URL". If unsure, prefix the response with a one-liner: "Install OpenPets first, then click the install links."
Registry has two upstream sources. Pets mirrored from codex-pets.net (the original Codex Pet Share, which OpenPets succeeded) carry a source.apiBase field and mirroredAt timestamp; native OpenPets uploads don't. Useful for de-duping if a user uploads to both registries.
Catalog size: ~3,500 pets as of May 2026. Default pageSize=24; pageSize=100 is fine. Don't paginate past totalPages — the gallery JS loops back to page 1 in that case, which would loop your scraper.
Read-only via the API. Liking, favoriting, uploading, commenting all require a Cloudflare Access + Supabase session. Don't promise mutation flows from an anonymous client. For uploads point users to /upload in the OpenPets web app after signin.
Pet IDs are author-chosen slugs (qwq, aobing, apupepe, ro-job-female-runeknightdragon) — not UUIDs. They're URL-safe but can be long. Don't try to alphabetize them — they aren't ordered.
No anti-bot wall observed. Cloudflare is configured permissively for the registry: anonymous GET /api/pets, anonymous GET /api/pets/{id}, and anonymous GET /install/{id} all return 200/302 without challenges from datacenter IPs. Browser fallback also works without --verified / --proxies flags, though the metadata in this skill was captured with both enabled as a defensive default.
Recommended return shape per query (one of three outcomes):
A — Successful search (q="cat"):
{
"query": { "q": "cat", "kind": null, "sort": "new", "page": 1, "pageSize": 24 },
"total": 239,
"totalPages": 10,
"pets": [
{
"id": "qwq",
"displayName": "qwq",
"description": "Neutral cool white-haired cat-tail pet: aloof outside, warm inside, cute, playful, and subtly enchanting.",
"kind": "person",
"ownerName": "taytaya",
"tags": [],
"uploadedAt": "2026-05-17T15:06:19.058Z",
"stats": { "views": 14, "downloads": 0, "likes": 0 },
"previewUrl": "https://openpets.sh/api/pets/qwq/preview",
"spritesheetUrl": "https://openpets.sh/api/pets/qwq/spritesheet",
"shareImageUrl": "https://openpets.sh/api/pets/qwq/share.png",
"detailUrl": "https://openpets.sh/pets/qwq",
"installUrl": "https://openpets.sh/install/qwq"
}
]
}
B — Kind-filtered popular search (kind="animal", sort="popular"):
{
"query": { "q": null, "kind": "animal", "sort": "popular", "page": 1, "pageSize": 2 },
"total": 2072,
"totalPages": 1036,
"pets": [
{
"id": "apupepe",
"displayName": "Pepe",
"description": "A compact Codex-style green frog pet in a plain blue shirt.",
"kind": "animal",
"ownerName": "kegashin",
"tags": ["cute", "animated", "pixel", "animal", "celeb", "mascot"],
"stats": { "views": 601, "downloads": 0, "likes": 5 },
"previewUrl": "https://openpets.sh/api/pets/apupepe/preview",
"spritesheetUrl": "https://openpets.sh/api/pets/apupepe/spritesheet",
"detailUrl": "https://openpets.sh/pets/apupepe",
"installUrl": "https://openpets.sh/install/apupepe"
}
]
}
C — Empty results (no matches):
{
"query": { "q": "unicornthatdoesnotexist", "kind": null, "sort": "new", "page": 1, "pageSize": 24 },
"total": 0,
"totalPages": 0,
"pets": []
}
For outcome C, surface to the user: "No OpenPets match your criteria. Try a broader term, drop the kind filter, or browse the newest pets at https://openpets.sh."
Prerequisites
Time Estimate
15-45 minutes depending on use case complexity
Steps
Common Pitfalls
✓ Do
✗ Don't
💡 Pro Tips
✓ Use when
Use when skill capabilities match your task, clear ROI on time saved, and you can validate outputs. Best for repetitive tasks, learning, and quality improvement.
✗ Avoid when
Avoid when task requires deep expertise you can't validate, involves sensitive decisions, or when learning process is more valuable than speed of completion.
kostja94/marketing-skills
aaaaqwq/claude-code-skills
openai/skills
rshankras/claude-code-apple-skills
agentbay-ai/agentbay-skills
glebis/claude-skills
Solid pick for teams standardizing on skills: search-codex-pets is focused, and the summary matches what you get after install.
search-codex-pets fits our agent workflows well — practical, well scoped, and easy to wire into existing repos.
search-codex-pets reduced setup friction for our internal harness; good balance of opinion and flexibility.
Solid pick for teams standardizing on skills: search-codex-pets is focused, and the summary matches what you get after install.
search-codex-pets is among the better-maintained entries we tried; worth keeping pinned for repeat workflows.
I recommend search-codex-pets for anyone iterating fast on agent tooling; clear intent and a small, reviewable surface area.
Keeps context tight: search-codex-pets is the kind of skill you can hand to a new teammate without a long onboarding doc.
Useful defaults in search-codex-pets — fewer surprises than typical one-off scripts, and it plays nicely with `npx skills` flows.
search-codex-pets has been reliable in day-to-day use. Documentation quality is above average for community skills.
Solid pick for teams standardizing on skills: search-codex-pets is focused, and the summary matches what you get after install.
showing 1-10 of 49