network-service-attacks

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides numerous command-line templates for standard networking and security tools, including nmap, netexec, snmpwalk, ldapsearch, and smbclient. These are intended to guide a security professional or an agent operating in pentest mode through service auditing.
  • [INDIRECT_PROMPT_INJECTION]: The skill involves processing untrusted data from network services (such as version strings and service banners) to determine subsequent actions (like searching for CVEs). This creates a theoretical attack surface where a malicious service could return crafted data to influence the agent.
  • Ingestion points: Service banners and version information retrieved from network scans.
  • Boundary markers: None present in the instructions.
  • Capability inventory: The skill utilizes tools capable of network interaction and file system access (via mounting shares).
  • Sanitization: No explicit sanitization or validation of the retrieved service strings is mentioned.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 11:01 PM
Security Audit — agent-trust-hub — network-service-attacks