The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides command templates for various security tools including nmap, sqlmap, nikto, subfinder, and curl.
Includes templates for active reconnaissance and exploitation, such as a directory traversal payload targeting /etc/passwd.
While these are provided for authorized testing, they represent a capability for active network interaction and probing.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it processes user-supplied target identifiers that are interpolated into command templates.
Ingestion points: Target URLs, IP addresses, and account names provided by the user during the interaction.
Boundary markers: The skill defines a mandatory confirmation gate requiring the user to state the exact target, confirm written authorization, and review the command impact before execution.
Capability inventory: Tools and techniques described include DNS enumeration, port scanning, web vulnerability scanning, SQL injection testing, and directory traversal.
Sanitization: No automated sanitization or escaping of the target inputs is described; the skill relies on manual user oversight and confirmation gates.

penetration-testing