Zenity Labs finds malicious AI agent skills in public registries

Researchers identified bad actor code in open repositories and released a security tool called AI Total.

Reported45 sec read

Aug 11: Zenity Labs finds malicious AI agent skills in public registries

Photo: Jefferson Santos · Unsplash

What happened

Zenity Labs found malicious AI agent skills inside public registries. These compromised skills target automated workflows when agents integrate them. Zenity Labs launched a new scanner called AI Total to inspect these agent capabilities.

The context

AI agents frequently download third-party tools to perform specialized tasks. Unvetted public registries introduce traditional supply-chain risks directly into automated AI systems.

Sources

Portal ERP ↗ via Google News Reported

See the full Board →