infrastructure-ops
Pass
Audited by Gen Agent Trust Hub on Jul 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is designed to facilitate remote infrastructure management by executing administrative commands such as service restarts (
systemctl), configuration edits (sed), and file inspection (cat) across multiple servers. - [PROMPT_INJECTION]: The skill defines patterns where the agent must read and interpret output ('screens') from remote hosts to make decisions. This introduces a surface for indirect prompt injection if a host displays malicious content designed to influence the agent's logic.
- Ingestion points: Remote terminal output retrieved using
read screenoperations inSKILL.md. - Boundary markers: No specific delimiters or instructions to ignore embedded commands in host output are provided.
- Capability inventory: Extensive capabilities including remote shell access (
wsh_send_input), session management, and system-level file modification. - Sanitization: No sanitization or validation of the content received from remote servers is mentioned.
Audit Metadata