A flaw in Red Hat Ansible Automation Platform 2 allows a user with permission to create or modify a project to supply a specially crafted SCM URL that is stored unchanged and passed directly to the git SCM module. The git command git ls-remote <scm_url> is invoked without a separator, so a URL such as "--upload-pack=<command>:x" is interpreted by git as the --upload-pack option. This option runs <command> under a shell and causes arbitrary code execution on the control‑plane task pod. The result of the command is returned via the project update stdout endpoint, giving the attacker a writable channel to exfiltrate data and enabling cross‑tenant compromise and lateral movement inside the cluster. The weakness is a classic OS command injection (CWE‑88) that affects confidentiality, integrity, and availability across all tenants that share the control plane.

The vulnerability impacts all installs of Red Hat Ansible Automation Platform 2. The affected product is the AutomationController component, which uses the Git SCM module to fetch project sources. No specific version constraints are listed beyond the platform major release due to the nature of the flaw.

The CVSS score of 9.9 indicates critical severity. EPSS data are not available, so no quantified likelihood of exploitation is currently known, but the lack of a mitigation path and the ability to execute commands via the control‑plane pod suggest a high risk in environments where users can create or modify projects. The vulnerability is not listed in CISA KEV. The attack vector is internal to the cluster: a compromised or malicious project configuration can be crafted by an authorized user and the malicious code will run with the privileges of the controller‑task pod. If exploited, the attacker can move laterally across tenants and modify or delete configuration data.