The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various security-focused command-line tools including ffuf for content discovery, arjun and paramspider for parameter discovery, and wpscan for WordPress vulnerability scanning.
[EXTERNAL_DOWNLOADS]: The instructions reference installing dependencies and wordlists from standard package repositories (apt, pip, gem) and the well-known SecLists repository on GitHub.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it ingests and processes untrusted data from target web applications through browser automation and tool outputs.
Ingestion points: Untrusted data enters the agent context through browser_get_page, browser_evaluate, and the output of scanning tools like ffuf and wpscan.
Boundary markers: The skill contains explicit instructions for the agent to "STOP" and return findings to the orchestrator immediately upon the confirmation of any injection point or vulnerability.
Capability inventory: The agent has capabilities for network communication, local file writing (specifically to the engagement/evidence/ directory), and state modification via a state MCP server (add_vuln, add_credential).
Sanitization: There are no explicit instructions for sanitizing or escaping the data retrieved from external web pages before it is processed by the LLM.

web-discovery