The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses local system utilities such as curl, nmap, wpscan, and msfconsole to perform discovery and penetration testing against remote targets. These commands incorporate user-supplied parameters, including target URLs, IP addresses, and wordlist file paths.
[DYNAMIC_EXECUTION]: The instructions provide templates for generating malicious PHP payloads at runtime. This includes a webshell implementation using system($_GET['cmd']) and a PHP-based reverse shell using exec to invoke an interactive bash shell directed to a remote IP and port.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external, untrusted web servers through automated scanning tools. This creates a surface where a malicious target could return crafted data—such as in HTML comments, HTTP headers, or REST API responses—intended to influence the agent's behavior.
Ingestion points: Command outputs from curl, wpscan vulnerability reports, and nmap script results when scanning external targets.
Boundary markers: The skill requires the agent to implement a mandatory confirmation gate where the user must verify the target, authorization scope, and exact command effects before any execution occurs.
Capability inventory: The agent has the ability to execute shell commands, manage local files (creating, zipping, and editing), and initiate network connections via penetration testing frameworks.
Sanitization: No explicit sanitization or filtering of target-derived data is defined; the skill relies on manual user review and confirmation of the agent's proposed actions.

wordpress-penetration-testing