careful

To add a safety guardrail, this Skill implements a pre-run command checker as a Bash hook that scans input for dangerous patterns like rm -rf and returns a permission decision to ask for confirmation before execution.

The override mechanism works by returning a permission decision to ask when a destructive command is detected, warning the user and providing an explicit path to proceed with the action if necessary.

Yes, you can use this in testing and production debugging scenarios to warn against destructive commands including kubectl delete and docker prune, preventing them from running unnoticed.

Yes, the command checker logs offenses for audit, providing audit-friendly records of attempted destructive actions that triggered the safety guardrails during your shell sessions.

A limitation of this pre-run command checker is that it scans for dangerous patterns in shell input, meaning it focuses on preventing destructive commands through warnings and optional overrides rather than blocking them entirely.