guard

Preventing accidental destructive bash commands during production debugging requires combining destructive command warnings with a directory-scoped edit boundary. This approach restricts edits outside a defined path and warns on risky shell actions.

Yes, you can restrict shell edits to a specific directory boundary by activating a safety mode. The system asks which directory edits should be restricted to, enforcing that edits outside the path are blocked.

Warning about risky commands during deployments is best handled by implementing PreToolUse hooks for Bash, Read, and AskUserQuestion. These hooks intercept destructive actions before execution and issue warnings.

To temporarily thaw guard mode for directory isolation, you reference sibling careful and freeze scripts. The freeze state is stored in the user's gstack state directory to manage the boundary scope.

Yes, full safety mode works with standard bash workflow security by adding an edit boundary layer. It enhances typical workflow security by actively preventing edits outside a restricted path rather than only issuing warnings.

Edit boundaries for shell commands should not be used when you need unrestricted write access across the entire file system. They are designed for prod debugging and deployment scenarios where limiting scope prevents catastrophic damage.