The vulnerability exists because the destination namespace for resource injection in Lighthouse is derived from an attacker-controlled label or annotation on the broker object without proper validation. This allows an attacker controlling a spoke cluster to inject unauthorized EndpointSlices and ServiceImports into any namespace on peer clusters, including critical system namespaces like kube-system and openshift-*. The lighthouse-agent service account uses this attacker-controlled label, enabling potential privilege escalation or service disruption across the cluster federation. The CVSS v3.1 score is 9.9 (critical), reflecting network attack vector, low complexity, low privileges required, no user interaction, changed scope, and high impacts on confidentiality, integrity, and availability.

Successful exploitation allows a remote attacker with control over a spoke cluster to inject unauthorized network resources into critical namespaces on peer clusters. This can lead to privilege escalation or other system compromises within the Kubernetes cluster federation, potentially disrupting cluster operations or enabling further attacks.

According to the Red Hat advisory, no effective mitigation is currently available that meets their criteria for ease of use, applicability, and stability. There is no official fix or patch released yet. Users should monitor the Red Hat advisory for updates and consider limiting exposure of spoke clusters to untrusted actors. Customers with a Red Hat Technical Account Manager (TAM) can review this CVE directly with their TAM for tailored guidance.