The Agentic Adversary: Speed, Scale, and Poisoned Toolchains – CrowdStrike Triple-T
Adversaries no longer need malware to break in. They log in with stolen identities, and they’re using AI to move at machine speed, automating reconnaissance, social engineering, and data theft. Meanwhile, the AI coding agents enterprises are adopting have opened a new way in: agents install their own packages, skills, and MCP servers with no human review. Drawing on front-line threat intelligence, this session shows how identity-based and AI-accelerated attacks unfold, traces a poisoned-toolchain attack from a single install to exfiltration, and outlines six practical points to stop it, plus three steps teams can start on Monday.
Adam Mitchell serves as a Platform Director at CrowdStrike, operating at the critical intersection
of threat intelligence and customer defense. Collaborating with the Incident Response and
Product teams, he translates frontline insights into actionable strategies for organizations under
attack. Adam demystifies sophisticated adversary tactics and helps customers leverage
CrowdStrike’s platform to address their unique risk landscape.
