The Agent Skills Directory

[PROMPT_INJECTION]: The skill documentation and evaluation files include common prompt injection keywords (e.g., 'ignore previous instructions', 'DAN'). These are contextually identified as detection targets for the auditing process and do not pose an active threat to the agent's own instruction set.
[COMMAND_EXECUTION]: The skill utilizes a local Python script ('scripts/audit.py') to perform deterministic security checks. It also executes basic filesystem discovery commands ('ls') to identify installed companion skills.
[EXTERNAL_DOWNLOADS]: The skill references the vendor's domain ('absolutelyskilled.pro') and repository for installing related security tools via the 'npx' command. These resources are consistent with the skill's stated source and functionality.

skill-audit