Vetted · Skills
BrowseHow vetting worksPublish
Sign inGet started
Vetted · Skills

A marketplace for AI agent skills where every skill is security scanned before it can be published.

Marketplace

  • Browse skills
  • Publish a skill
  • Your dashboard

Trust

  • How vetting works
  • Open source (MIT)

Built as an open teaching project. Free to use and build on.

Automated vetting reduces risk. It is not a guarantee of safety.

Scanned before it ships

Publish a skill

Paste your SKILL.md, upload a bundle, or import a public repo. Every submission runs through four independent scanner layers before it can reach the catalog. A single critical finding blocks publication, so the trust badge on a listed skill actually means something.

Trying the scanner? Paste a snippet with curl ... | bash or the phrase “ignore previous instructions” and watch it get blocked.

A critical finding blocks publication. Nothing goes live until the scan passes.

What we check

Four layers read every submission. Bundled scripts are analyzed statically and never executed.

  • Hidden-text detection

    01

    Decodes invisible Unicode, bidi, and zero-width characters that smuggle instructions past a human reviewer.

  • Dangerous-instruction patterns

    02

    Flags prompt-override phrasing, secret/credential reads, and network exfiltration in the SKILL.md.

  • Static script analysis

    03

    Inspects bundled scripts for remote code execution, subprocess calls, and obfuscation. Never runs them.

  • LLM injection classifier

    04

    A Claude model reads the skill and rates prompt-injection risk. A triage signal, not the final gate.