Agentic AI security research, mental models, and practitioner tools. Building in public — the Kill Chain is live, the rest is in progress.
Understanding how agentic AI systems get attacked. Attack vectors, threat models, and emerging research.
What to implement. Guardrails, sandboxing, least privilege, and enforcement patterns for agent systems.
How to test. Red-teaming methodology, eval suites, and evidence-first auditing for agent systems.
Runtime security. Behavioral baselines, logging, and anomaly detection for agents in production.
For security leaders. Risk frameworks, compliance mapping, and governance patterns for agentic AI.
Secure patterns for people building agents. MCP design, auth, tool scoping — code-level guidance.
Research and practitioner writing on agentic AI security. One piece live, more in progress.