Episode 17 | Agents Attacking and How App Control Helps

Everyone's worried about agents escaping sandboxes and hacking governments. But is that the real threat? In this episode, Jose and Mike go through what the research is actually showing. RussianPanda's breakdown of ransomware crews already using skills files with secondary payloads, the Chinese CERT flagging skills as an active threat vector, and a SNCC analysis finding 36% of analyzed skills contained malicious payloads. They also walk through LOL Skills live, scanning OpenClaw, Anthropic's marketplace, OpenAI's store, and GitHub,over 26,000 results after about a month of scanning. The parting question: is any of this actually preventable today? ⚔️ LOLSkills: https://lolskills.io 🔗 MagicSword is free for up to 100 endpoints, get started here: https://www.magicsword.io/dashboard/s... 🚀 Ready for More? Go Enterprise. Enterprise unlocks unlimited endpoints, 90+ days of data retention, and a much deeper feature set. Book a demo: https://www.magicsword.io/book-demo #AIAgents #LLMSecurity #SkillsSecurity #LOLSkills #PreventionLabLive #CyberSecurity #AppControl