Tool updates: lots of security and logic fixes, (Mon, Mar 23rd)

The author used the AI assistant Claude to review several of their public Python scripts and found multiple security issues — including a logic-inversion bug, TOCTOU race condition, ambiguous hash-handling, overly permissive file permissions and possible symlink attacks, encoding problems, and a potential mail header injection — some of which run as root and could be leveraged for privilege escalation. The piece emphasizes integrating AI code review into routine practice but does not report any active exploitation or breach.