Kubernetes is widely adopted for orchestrating cloud-native workloads, but its complexity makes it challenging to build test environments that are both realistic and repeatable for cybersecurity experimentation. Security assessments and red/blue team training often rely on either overly synthetic labs or time-consuming manual setups, limiting systematic evaluation of misconfigurations, attack paths, and defensive controls, and making it harder to validate observability-driven workflows for CTI generation. The proposed solution provides a reproducible testbed for studying attacker behaviour and validating defensive strategies and observability capabilities in Kubernetes environments, an area that remains only partially covered by existing security training and benchmarking solutions. This thesis describes a containerized laboratory platform that automatically provisions an isolated environment with a multi-node Kubernetes cluster and realistic vulnerable microservice workloads, including control-plane and policy-layer issues such as RBAC misconfigurations, overly permissive service accounts, and insecure access to core components.