network-service-attacks
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions involve processing untrusted data from network service banners and enumeration tool outputs (nmap, snmpwalk, ldapsearch) which are then used to inform subsequent command execution (netexec, smbclient). * Ingestion points: Output from network discovery and enumeration tools in SKILL.md. * Boundary markers: No explicit instructions to the agent to sanitize or ignore instructions embedded in service banners or responses. * Capability inventory: Shell command execution via tools like netexec, enum4linux-ng, smbclient, snmpwalk, showmount, and nmap. * Sanitization: No sanitization logic for service-returned data is provided.
Audit Metadata