OperTraitors: How Kubernetes Operators Betray Your Security Posture
Kubernetes operators frequently receive overly permissive wildcard RBAC permissions that transform them into silent backdoors exploitable through supply chain or dependency compromises. The research introduces OperTraitor, an LLM-powered auditing tool that scans operator manifests across OperatorHub and local deployments, calculating risk scores by comparing documented functionality against actual granted privileges. The emerging threat of agentic AI operators (those using LLMs for autonomous cluster management) amplifies this risk significantly—external AI agents or LLM-enhanced logic inheriting broad RBAC can read sensitive data and execute unintended cluster-wide actions. The authors discovered a CVSS 8.8 vulnerability (CVE-2026-6389) in IBM’s Turbonomic platform through this methodology, and provide practical guidance for downscoping service accounts to mitigate similar risks in production environments.