TitanCA: Lessons from Orchestrating LLM Agents to Discover 100+ CVEs
The episode discusses the 'TitanCA' paper, a methodology for discovering vulnerabilities (CVE's) using orchestrated LLM agents. The hosts analyze how TitanCA uses multi-layered, structured reasoning to ensure accuracy and auditability. Key improvements discussed include domain adaptation for enterprise code and prioritizing cost-aware pipeline ordering to create a reliable, scalable security tool.