A flaw affecting four leading AI coding agents could have let attackers slip malicious code into skills after users had already downloaded them. The vulnerability highlights immediate security risks as workers rely more heavily on AI agents and third-party tools. Full story: https://thein.fo/4dnoJd4