The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches the Nuclei scanner binary from the official ProjectDiscovery GitHub repository and pulls the OpenVAS scanner container from Greenbone's Docker Hub. These are recognized, well-known providers in the cybersecurity community.
[COMMAND_EXECUTION]: Instructions include shell commands for running Docker containers, administering scans via the OpenVAS Management Protocol (OMP), and executing Nuclei and OpenSCAP scans. It also provides a Python implementation for interacting with the Nessus API.
[INDIRECT_PROMPT_INJECTION]: As a security tool, this skill inherently processes untrusted data from target systems (e.g., HTTP headers, service banners, and vulnerability scan results).
Ingestion points: Scan result files such as nuclei-results.txt, results.xml, and report.html (referenced in SKILL.md).
Boundary markers: The skill features a strong safety mechanism—a mandatory confirmation gate—which requires the agent to verify authorization, state the scope, and show exact commands for user approval before any active probes.
Capability inventory: Shell command execution (via Docker, OMP, Nuclei, OSCAP) and network requests via the Python requests library.
Sanitization: While explicit character escaping is not shown in the templates, the system relies on human-in-the-loop validation and the platform's default safety guardrails.

vulnerability-scanning