The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill documents workflows that ingest large amounts of untrusted data from external infrastructure (DNS records, search engine results, and third-party APIs) and pipes this data into downstream tools such as httpx, nmap, and eyewitness.
Ingestion points: SKILL.md and references/config.md define processes where amass and theHarvester collect external subdomain names and infrastructure data.
Boundary markers: No specific delimiters or "ignore instructions" warnings are present in the documentation for the data pipeline.
Capability inventory: The skill utilizes shell execution for several reconnaissance and mapping tools (amass, dnsx, httpx, nmap, theHarvester, eyewitness).
Sanitization: The skill relies on the individual tools' internal sanitization of external strings; however, this is standard behavior for security-focused reconnaissance workflows.
[COMMAND_EXECUTION]: The skill provides numerous examples and a bash script template for executing security tools via the command line. These commands are intended for authorized security testing and infrastructure discovery within a laboratory or professional environment.

amass