sysadmin-monitor

Pass

Audited by Gen Agent Trust Hub on Aug 3, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to use standard Linux system utilities, including free, df, systemctl, and journalctl, to gather diagnostic information about system resources and service status.
  • [EXTERNAL_DOWNLOADS]: The instructions include a reference to the author's GitHub repository at github.com/HermeticOrmus/linux-sysadmin-skills for supplemental documentation.
  • [PROMPT_INJECTION]: The skill establishes an indirect prompt injection surface by processing system-generated data.
  • Ingestion points: System logs accessed via journalctl and process information from top or similar tools.
  • Boundary markers: Absent; there are no explicit instructions to the agent to disregard embedded commands in the log output.
  • Capability inventory: The agent uses shell command execution to retrieve diagnostic data.
  • Sanitization: Not specified in the instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 3, 2026, 04:06 AM
Security Audit — agent-trust-hub — sysadmin-monitor