A flaw in the community.proxmox Ansible collection’s proxmox_pct_remote connection plugin allows unquoted string concatenation to build a pct exec command, which is then passed to the Proxmox host’s shell via Paramiko. Shell metacharacters, such as '>', '|', ';', '&&', or '$()', are interpreted on the hypervisor host rather than inside the target LXC container, enabling commands intended for an unprivileged container to run as root on the hypervisor. This container‑to‑host trust boundary breach constitutes remote code execution on the Proxmox host.

The vulnerability affects all deployments using the community.proxmox Ansible collection that employ the proxmox_pct_remote connection plugin. Specific product or version details were not provided, so any version of the collection with the vulnerable exec path is susceptible. Operators of Proxmox environments that rely on this plugin for LXC container management should verify their collection version and downgrade or patch accordingly.

The CVSS score of 9.9 indicates critical severity, while the EPSS score is currently unavailable, leaving the exploit probability unclear but not negligible. The vulnerability is not listed in the CISA KEV catalog. The likely attack vector is an Ansible controller that uses the proxmox_pct_remote plugin to run pct exec commands; an attacker with the ability to influence the playbook or inventory can execute arbitrary shell commands on the Proxmox host as root.