The Agent Skills Directory

[COMMAND_EXECUTION] (HIGH): The skill is designed to execute powerful system commands with sudo privileges across SSH. These include package list updates (apt update), firewall configuration (ufw), and service status checks. While configured for 'read-only' by default, the underlying capability allows for full system modification.
[CREDENTIALS_UNSAFE] (HIGH): The skill requires the user to provide sensitive SSH access details and sudo permissions. Granting an AI agent root-level access to a server is a high-risk operation that could lead to full system compromise if the agent's logic is subverted.
[PROMPT_INJECTION] (HIGH): The skill is highly vulnerable to Indirect Prompt Injection because it ingests untrusted data from the VPS environment and processes it via the LLM.
Ingestion points: System logs via journalctl, container metadata via docker ps, and package lists via apt list in SKILL.md.
Boundary markers: Absent. The instructions do not specify any delimiters or warnings to ignore instructions embedded within the command outputs.
Capability inventory: SSH network access, sudo privilege escalation, and file system modification via package managers as described in the Workflow section.
Sanitization: Absent. There is no mention of filtering or sanitizing the data returned from the VPS before it is parsed by the agent.

vps-checkup