linux-troubleshooting
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from system logs, kernel messages, and process outputs, which represents a potential attack surface.
- Ingestion points: System logs (journalctl, /var/log/syslog), kernel messages (dmesg), and process listings (ps, top) mentioned in SKILL.md.
- Boundary markers: Absent; instructions do not specify delimiters for data processed by the agent.
- Capability inventory: The skill includes commands for service management (systemctl restart), process tracing (strace), and network connectivity testing (curl) in SKILL.md.
- Sanitization: The skill provides manual guidance to redact credentials and personal data before sharing results.
- [COMMAND_EXECUTION]: The workflow involves the execution of various administrative and diagnostic shell commands.
- Evidence: Standard Linux commands for resource monitoring, process management, and networking are listed throughout the workflow phases in SKILL.md.
Audit Metadata