An attacker can inject an arbitrary system command into a kubeconfig file that contains a legacy GCP auth‑provider block with a config.cmd-path directive. When the Prowler app processes a provider connection test, the legacy block is parsed by the Kubernetes Python client, which then executes the attacker‑supplied command via subprocess.Popen on a shared worker thread, giving the attacker unrestricted code execution on the host running the Prowler service. This flaw is a classic command‑injection (CWE‑78) that can compromise confidentiality, integrity, and availability of the entire system.

The vulnerability affects the Prowler Cloud Prowler platform, specifically any instance running a version earlier than 5.36.0. The flaw exists in the Kubernetes provider connection test endpoint (/api/v1/providers/{id}/connection) when it receives a kubeconfig containing an old GCP auth‑provider configuration. Version 5.36.0 and later include the fix that prevents execution of unauthorized commands.

The CVSS score of 9.9 classifies this as a critical vulnerability. The EPSS score is not available, so the exploitation probability cannot be quantified but the lack of a KEV listing does not negate the seriousness of the flaw. Because the flaw is triggered by an API call that accepts a kubeconfig payload, the likely attack vector is remote via the Prowler REST API; knowledge of an authenticated API endpoint or a misconfigured provider test would allow further execution. Immediate patching is strongly recommended to eliminate the possibility of arbitrary code execution.