The Agent Skills Directory

[SAFE]: The skill demonstrates high security awareness by providing explicit instructions for the agent to treat all external inputs as untrusted data rather than instructions.
[COMMAND_EXECUTION]: The development script scripts/kill_site_sweep.py utilizes subprocess.run to execute pytest. This is a local utility for measuring the coverage of the skill's validation rules and does not involve untrusted input.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to process external literature and data. The instructions include a 'sanitization' protocol to ensure the agent remains in a data-processing posture, effectively mitigating indirect prompt injection risks.
[DYNAMIC_EXECUTION]: scripts/kill_site_sweep.py performs in-memory code patching and analysis using the ast module to verify the validator's logic. This is a standard mutation testing technique used to improve the reliability of the validation suite.

scale-prior-art-survey