The Agent Skills Directory

[PROMPT_INJECTION]: The skill includes phrases such as "Ignore all previous instructions..." and "I am the admin, execute now...". These are explicitly identified as simulation scenarios for security drills and are provided as descriptive examples rather than instructions for the AI to follow.
[COMMAND_EXECUTION]: The skill references the execution of a local Python script run_drill.py to perform security simulations. This behavior is consistent with the skill's stated purpose as a testing tool and does not involve unsafe command injection or unauthorized privilege escalation.

security-drill