/

CVE-2026-73263

Severity:

CRITICAL RISK

(9.9)

OS Command Injection (CWE-78) in Prowler K8s Provider

CVE-2026-73263

Awaiting Analysis

Vulnerability type

Category:

OS Command Injection

CWE:

CWE-78

Impact:

Remote Code Execution

Severity:

CRITICAL RISK

(9.9)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

A command injection vulnerability in Prowler's Kubernetes provider allows attacker-controlled kubeconfig authentication settings to invoke operating system commands on a shared worker.

The issue becomes risky when an attacker can submit or modify provider configuration containing legacy GCP authentication fields that are processed during a connection test.

Affected context

The vulnerability affects Prowler deployments using the Kubernetes provider before the relevant security fix, where provider connection testing accepts kubeconfig content and loads it through the Kubernetes Python client. The exposure depends on whether an untrusted user or tenant can provide Kubernetes configuration that reaches the legacy GCP authentication provider path on a shared worker.

Exploit preconditions

  • Attack surface: Prowler's Kubernetes provider connection-test API
  • Required access: Ability to submit or influence a provider configuration containing kubeconfig content
  • Trigger point: Loading the kubeconfig during a provider connection test
  • Why exploitation is possible: Validation checks recognize only standard exec blocks, while legacy GCP authentication settings can still cause the Kubernetes client to invoke a configured command through a subprocess

Why this happens

The vulnerability results from incomplete validation of executable authentication configuration in kubeconfig data. The application checks for one supported command-execution representation but does not reject equivalent legacy GCP auth-provider settings before passing the configuration to a client library.

During parsing and authentication, the Kubernetes Python client can interpret attacker-controlled command fields and launch them through a subprocess. This behavior maps to CWE-78: Improper Neutralization of Special Elements used in an OS Command, because untrusted configuration reaches an operating-system command execution sink.

Potential impact

  • Arbitrary commands may execute with the privileges of the shared worker
  • Credentials, tokens, or cloud metadata accessible to the worker could be exposed
  • Other tenants, scans, or workloads on the worker could be affected
  • The Prowler service or connected cloud environments could be compromised
  • Attackers could establish persistence or conduct lateral movement

General mitigation guidance

Mitigation generally focuses on treating kubeconfig content as untrusted data and preventing any configuration-controlled command execution.

Common defensive patterns include:

  • Rejecting all executable authentication mechanisms unless explicitly required and tightly allowlisted
  • Detecting both current and legacy command-bearing kubeconfig formats
  • Using non-executable, short-lived authentication mechanisms where feasible
  • Applying strict schema and semantic validation before client-library loading
  • Isolating connection-test workers and restricting their filesystem, network, and cloud permissions
  • Enforcing tenant separation and least privilege for shared workers
  • Preventing untrusted users from supplying arbitrary provider configuration

How teams detect issues like this

Organizations can look for this class of issue through:

  • Static analysis and code review rules for subprocess APIs reached from configuration or request data
  • Searches for kubeconfig parsing paths that accept exec or legacy auth-provider command fields
  • Tests covering alternate configuration representations and parser-library behavior
  • API policy checks that restrict provider configuration fields and executable authentication methods
  • Runtime monitoring for unexpected child processes spawned by connection-test workers
  • Audit correlation between provider-configuration submissions and process execution events

Automated security review tools may help identify incomplete validation and command-execution data flows.

Sources

SourceLink
github.comgithub.com/prowler-cloud/prowler/commit/0b782fcb8c24ece7bd38deede2b8f13d8583e39c
github.comgithub.com/prowler-cloud/prowler/pull/12091
github.comgithub.com/prowler-cloud/prowler/releases/tag/5.36.0
github.comgithub.com/prowler-cloud/prowler/security/advisories/GHSA-ccqh-6cjc-wp4j

Frequently asked

questions

It can be exploitable through the provider connection-test API when an attacker can submit kubeconfig content that reaches the vulnerable authentication path. The practical scope depends on API exposure and authorization controls.

The attacker generally needs authenticated or otherwise authorized access to create or modify a Kubernetes provider configuration, or another means of influencing the submitted kubeconfig content.

Treat kubeconfig input as untrusted, reject command-bearing authentication configurations—including legacy forms—and use non-executable authentication with strict validation before client-library processing.

Validate every equivalent representation of dangerous configuration, apply least privilege and worker isolation, and test security controls against parser and library behaviors rather than relying on a single syntax check.

Find out what's already
exploitable in your codebase.