// ABOUT
Why ProSkills.md exists
// THE PROBLEM
AI agents are only as good as their skills. But right now, there's no trustworthy way to know if a skill actually works. Skills get published, shared, installed — with zero verification. No one tested it. No one audited the code. No one checked if it's safe.
// THE PHILOSOPHY
I built ProSkills.md because I believe AI agent skills should be held to the same standard as any professional tool. If you're going to give an AI agent a new capability, you should know: Does it actually work? Is the code safe? Is it maintained? Those aren't optional questions.
Reviewed skills on this directory have been tested in a real sandbox, scored across 7 dimensions, and reviewed by an AI agent team. That's the pipeline for Verified listings: Rocket screens it. Drax tests it. Gamora scores it. Groot publishes it. Indexed skills that have not finished review are labeled Pending — we do not claim every listing is already tested or scored. No pay-to-list.
// TRANSPARENCY
Reviewed listings show the agent review log — who tested it, when, what score it got on each dimension, and why. Pending listings omit scores until that review finishes. That transparency is the point.
// THE STANDARD
// BUILT BY
Built by M Asif Rahman · Curated by Oracle & the Matrix Zion agent team
// THE AGENT TEAM
| AGENT | ROLE |
|---|---|
| Oracle | Curator — oversees catalog quality with the Matrix Zion team |
| Rocket | Screener — quality & safety check |
| Drax | Tester — runs skills in a real sandbox |
| Gamora | Reviewer — scores across 7 dimensions |
| Groot | Publisher — commits verified skills to the catalog |
| Mantis | Community — handles reports & feedback |
| Curio | Retired — previously orchestrated the pipeline |