The Agent Skills Directory

[PROMPT_INJECTION]: No malicious instructions or attempts to bypass agent safety guidelines were detected. The instructional language used is appropriate for defining the evaluator role and evaluation criteria.
[DATA_EXFILTRATION]: No patterns for accessing sensitive files (e.g., credentials, ssh keys) or exfiltrating data to external domains were identified.
[EXTERNAL_DOWNLOADS]: No unauthorized package installations or remote script downloads were found. Code examples use standard libraries for demonstration purposes.
[COMMAND_EXECUTION]: No dangerous shell command executions, privilege escalation attempts, or persistence mechanisms were detected.
[REMOTE_CODE_EXECUTION]: The skill does not contain any remote code execution or dynamic code evaluation patterns.

advanced-evaluation