The Agent Skills Directory

Privilege Escalation (MEDIUM): The skill utilizes sudo for numerous commands in references/ubuntu-docker-checkup-commands.md to check system status (e.g., UFW, fail2ban, journalctl). While this is required for a system-level audit, it grants the agent elevated permissions on the target host. Severity is downgraded from HIGH as it is essential for the primary skill purpose.
Data Exposure & Exfiltration (MEDIUM): The skill reads sensitive local and remote configuration files, including ~/.ssh/config and /etc/ssh/sshd_config. Access to these files is a high-risk activity as they contain security-critical configurations.
Indirect Prompt Injection (LOW): The skill processes untrusted output from various system tools. Evidence Chain: 1. Ingestion points: Command outputs from journalctl, docker ps, and apt list as seen in references/ubuntu-docker-checkup-commands.md. 2. Boundary markers: None. 3. Capability inventory: Full shell access and sudo capabilities on the remote system. 4. Sanitization: Absent.

vps-checkup