The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides documentation for executing various shell commands involving powerful security tools like wpscan, curl, and jq. These commands are designed to interact with external WordPress installations for reconnaissance and vulnerability assessment.
[DATA_EXFILTRATION]: The skill documents the use of curl and wpscan to perform network requests to external domains. While these are intended for security testing, they involve interacting with non-whitelisted domains to fetch or probe data (e.g., checking for exposed .log or .bak files).
[INDIRECT_PROMPT_INJECTION]: The skill outlines workflows where the agent ingests data from external targets (e.g., scan results in JSON format or user lists) and processes it. This creates a surface where malicious content embedded in a WordPress target (like a crafted plugin name or user metadata) could potentially influence the agent's logic or downstream commands.
Ingestion points: scan.json, found_users.txt, users.json, and direct outputs from curl requests to WordPress endpoints.
Boundary markers: No boundary markers or 'ignore' instructions are provided to delimit untrusted content.
Capability inventory: The agent has access to wpscan, curl, jq, and file-writing capabilities via shell redirection.
Sanitization: No sanitization or validation steps are documented before the ingested data is used in subsequent command-line arguments.

wpscan