The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documentation includes snippets demonstrating command injection vulnerabilities, such as rm -rf / and system("grep -r '$pattern' ..."). These are clearly marked as "Bad" or "Anti-Patterns" to educate users on how to prevent such attacks through safe practices like list-form system() and three-argument open().
[SQL_INJECTION]: The skill provides examples of SQL injection (e.g., "SELECT * FROM users WHERE email = '$email'") as part of its educational content, contrasted with safe parameterized query examples using DBI placeholders.
[REMOTE_CODE_EXECUTION]: The skill identifies the use of eval with user input as a critical vulnerability to be avoided, following standard security best practices.
[EXTERNAL_DOWNLOADS]: The documentation references standard and well-known Perl modules available via CPAN (e.g., DBI, IPC::Run3, JSON::MaybeXS). These are industry-standard libraries for the Perl ecosystem and do not represent a security risk.

perl-security