Pentest Commands
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHCOMMAND_EXECUTION
Full Analysis
- Data Exposure (HIGH): The skill provides command templates targeting highly sensitive files. Evidence: The command 'ssh2john id_rsa' in SKILL.md involves processing private SSH keys for credential cracking.
- Indirect Prompt Injection (HIGH): The skill facilitates the ingestion of external data from untrusted targets while providing the agent with powerful side-effect capabilities.
- Ingestion points: Commands for 'nmap', 'nikto', 'sqlmap', and 'tshark' in SKILL.md are designed to ingest data from potentially hostile network services into the agent's context.
- Boundary markers: No delimiters or protective instructions are used to prevent the agent from obeying instructions embedded in the output of these tools.
- Capability inventory: The skill provides commands for Metasploit exploits, reverse shell generation via 'msfvenom', and remote command execution via 'sqlmap --os-shell'.
- Sanitization: There is no evidence of sanitization or validation for the data processed by these commands.
- Dual-Use Capability (INFO): The skill is a comprehensive reference for offensive tools which, while intended for pentesting, can be used for unauthorized system compromise.
Recommendations
- AI detected serious security threats
Audit Metadata