The Agent Skills Directory

[SAFE]: The skill serves as an educational resource for secure Perl programming. It provides 'Good' vs 'Bad' code examples to illustrate security concepts like taint mode, parameterized queries, and safe system calls.
[SAFE]: The destructive system commands (e.g., rm -rf /) flagged by static analysis are clearly labeled as examples of malicious payloads that could be injected into vulnerable code, rather than instructions to be executed by the agent.
[COMMAND_EXECUTION]: The skill suggests using the perlcritic tool for security auditing and demonstrates safe methods for calling external processes using list-form system() calls to avoid shell injection.
[INDIRECT_PROMPT_INJECTION]: While the skill is designed to review potentially untrusted Perl code for security vulnerabilities (an ingestion point for external data), it provides explicit safety boundaries by teaching how to untaint inputs, validate paths, and use delimiters to separate code from data.

perl-security