network-appliance-attacks

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to use common security auditing tools including ike-scan and openssl to evaluate the security posture of perimeter appliances and VPN configurations.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and interpret data from external network appliances, such as service banners, NTLM Type-2 challenges, and SSL certificate information, which could potentially contain malicious payloads if the target is controlled by an adversary.
  • Ingestion points: Network responses from external targets (banners, certificates, NTLM Type-2 challenges) described in SKILL.md.
  • Boundary markers: No specific delimiters or boundary markers are defined for handling the ingested network data.
  • Capability inventory: The skill utilizes tools like ike-scan, nmap, and openssl to interact with and collect data from network targets.
  • Sanitization: There are no instructions for sanitizing or escaping content retrieved from the network services before analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 11:01 PM
Security Audit — agent-trust-hub — network-appliance-attacks