proxmox-auth
Warn
Audited by Gen Agent Trust Hub on Aug 29, 2026
Risk Level: MEDIUMDATA_EXFILTRATIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONPRIVILEGE_ESCALATION
Full Analysis
- [DATA_EXFILTRATION]: The skill provides instructions to access and read sensitive Proxmox cluster and container configuration files, including /etc/pve/corosync.conf and LXC configurations in /etc/pve/nodes/. These files contain sensitive infrastructure metadata, network topology, and system configuration details.\n- [COMMAND_EXECUTION]: The skill facilitates the execution of arbitrary commands on Proxmox host nodes via SSH and within LXC containers using the 'pct exec' tool. This includes high-impact actions such as rebooting the server, managing services, and performing system-level restores.\n- [PRIVILEGE_ESCALATION]: The skill instructions explicitly encourage the use of root-level access via SSH ('ssh root@') for administrative tasks on the cluster nodes and within containers, bypassing standard user permission constraints.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external data from system logs and configurations that could potentially contain malicious instructions, creating a vulnerability surface.\n
- Ingestion points: The agent is instructed to read system logs using 'journalctl' and examine local system configuration files from the Proxmox cluster filesystem.\n
- Boundary markers: There are no explicit instructions or delimiters used to separate untrusted system data (logs/configs) from the agent's core instructions.\n
- Capability inventory: The skill possesses extensive capabilities, including shell command execution on host nodes and within containers, along with the ability to read system-level configuration files.\n
- Sanitization: The skill does not mention or implement any sanitization or validation of the log content or configuration data before it is processed by the agent.
Audit Metadata