guard

To prevent destructive commands during debugging, you need a safety mode that warns before executing high-impact actions like rm -rf or DROP TABLE. This skill intercepts and warns against destructive patterns using pre-tool checks before execution.

To restrict file edits to a specific directory, you establish an edit boundary that limits writes to a chosen path. This skill applies freeze boundary checks via hook scripts to block any file modifications outside your selected directory.

Preparing production-safe edits requires combining directory-scoped restrictions with destructive command warnings. This skill enforces both by running pre-tool checks for dangerous patterns and applying boundary checks on write actions to protect live-adjacent systems.

Destructive command warnings in interactive development work by running pre-tool checks that identify high-impact patterns like force-push or DROP TABLE. The system intercepts these operations and alerts you before the action is executed, preventing harmful changes.

Yes, you can use directory-scoped edit restrictions alongside standard debugging workflows. The safety mode applies boundary checks specifically to edit and write actions via hook scripts, allowing unrestricted work inside the safe workspace while blocking external modifications.

The limitation of using edit boundaries for production safeguards is that restrictions apply only to file edits and write actions within the configured path. You must manually activate guard mode and provide an absolute directory path to enforce the boundary checks.