find-skills — Meta-skill for Discovering Other Skills
find-skills is the #1 install on skills.sh (1.8–2.1M installs) and the standard starting point for anyone new to the skills ecosystem. It adds discovery, search, and install commands to any compatible agent.

find-skills
find-skills is the meta-skill: a skill that helps you find, evaluate, and install other skills. It is the most-installed skill on skills.sh at roughly 1.8M–2.1M installs and is published by Vercel Labs as part of the npx skills CLI distribution. Install it first, use it to discover the rest of the catalog here.
When to use
- You're new to the skills ecosystem and want a guided entry point
- You want to add search/discovery commands to your agent (
/find-skills <query>) - You're curating a private set of vetted skills and want a known-good starting install
When not to use
- You're already comfortable browsing agentskills.io and skills.sh directly —
find-skillsis convenience, not a gate - You have a strict allowlist and don't want a network-discovering tool in the agent's instruction set
How it works
find-skills ships as a SKILL.md with a small set of progressive-disclosure references. At install time, the agent's skill catalog gets the find-skills entry with its name and description (~100 tokens). When you call it, the agent loads the full instructions and can drive the npx skills CLI to search, install, and report on skills without you leaving the chat.
Install
npx skills add vercel-labs/skills
Full SKILL.md
Full source at the canonical repo: github.com/vercel-labs/skills. Embedded copy omitted at launch — the repo is MIT-licensed but the canonical SKILL.md is updated frequently; the link above always tracks the latest vetted copy.
Verify it yourself
Note:
Even MIT-licensed, high-installation skills are not inert. Run the verification block before installing in any environment with real credentials.
Quick static check — clone to a scratch directory and grep for: curl | sh, base64 -d, process.env, requests.post, ignore previous, and any line in the SKILL.md that asks the agent to hide output from you. Any hit → re-vet with the two-model review prompt.
Two-model review prompt — paste the full SKILL.md into two different models with the prompt from the Vetting Guide. Confirm both return LOW or MEDIUM with no injection class hit.
Review transparency
| Field | Value |
|---|---|
| Last reviewed | 2026-08-28 (launch baseline) |
| Reviewers | site baseline (static scan only at launch) |
| Source repo | github.com/vercel-labs/skills |
| License | MIT |
| Install signal | 1.8M–2.1M installs (skills.sh) |
| Re-vet trigger | any upstream SKILL.md change; >30 days since review |
Related
Related Articles & Guides
#skillsskill-creator — Build Your Own Agent Skills From Recorded Workflows
skill-creator is Anthropic's meta-skill for turning a recorded session or a procedure into a vetted, redistributable SKILL.md. The bridge between a one-off prompt workflow and a durable skill in the catalog.
#skillsSkills Collection — Curated Agent Skills for Every Tool
A curated, vetted catalog of agent skills (SKILL.md) that work across Claude Code, OpenCode, Gemini CLI, Codex, Cursor, and 40+ other agents. Each skill includes install tabs, full source, and instructions for verifying it yourself.
#skillsHow to Vet an Agent Skill With Models
A repeatable methodology for vetting agent skills before installing. Static red-flag scan plus a copyable two-model review prompt that flags prompt injection, exfiltration, and over-broad permissions. Built on the Aug 2026 security research wave.