CyberProof found malicious skills in public marketplaces that turn AI coding agents into insider threats, executing shell scripts without user consent.
CyberProof found malicious skills in public marketplaces that turn AI coding agents into insider threats, executing shell scripts without user consent.