The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions in SKILL.md for the agent to execute a variety of command-line security tools, such as subfinder, ksubdomain, naabu, httpx, nuclei, spray, and ffuf, as well as curl for interacting with external APIs.
[EXTERNAL_DOWNLOADS]: The methodology in SKILL.md relies on external security tools and references third-party API services (FOFA, Quake, Hunter) for information gathering. Although no direct installation scripts are included, the skill is dependent on these external resources.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from untrusted external environments which could potentially contain adversarial instructions targeting the agent.
Ingestion points: Data enters the agent context from OSINT search engines (FOFA, Quake, Hunter), SSL certificate logs, public code repositories (GitHub/GitLab), and active network scanning results, as described in SKILL.md.
Boundary markers: The instructions in SKILL.md do not include guidance on using delimiters or protective prompts to isolate untrusted data from the agent's core instructions.
Capability inventory: SKILL.md provides a roadmap for executing network probes, vulnerability scans, and exploitation attempts across assets using multiple CLI tools.
Sanitization: There is no mention of sanitizing or validating the content retrieved from external targets before the agent interprets or acts upon it.

red-team-assessment