tcp-ip
Warn
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: MEDIUMCOMMAND_EXECUTION
Full Analysis
- [Command Execution] (MEDIUM): The skill provides instructions for using high-impact system utilities that typically require root privileges.
- Evidence: Commands such as
tcpdump,iptables, andfirewall-cmdmodify system state or access raw network devices, requiring elevated permissions. - [Indirect Prompt Injection] (MEDIUM): The skill describes operations that process untrusted data from external network sources.
- Ingestion points: Raw network packets via
tcpdump, and DNS records viadig,nslookup, andhost(File: SKILL.md). - Boundary markers: Absent. No instructions are provided to delimit or sanitize data received from the network before the agent interprets it.
- Capability inventory: The skill provides the agent with the ability to modify network configurations (
iptables,ip route), perform network probes (nc,telnet), and capture traffic. - Sanitization: Absent. The skill does not provide methods for filtering or validating external inputs.
Audit Metadata