careful

To prevent destructive shell commands like rm -rf from executing, you can use a guardrail tool that intercepts risky patterns via a PreToolUse hook. It detects dangerous operations and prompts for confirmation before allowing the shell session to proceed, stopping accidental data loss.

Yes, you can add a confirmation prompt for git push --force and kubectl delete in production environments. This Skill detects these specific risky patterns in your shell sessions and automation workflows, prompting you to review risks and choose to proceed or cancel.

The best way to guard against destructive commands across development and production environments is applying a prompt-based decision logic tool. It intercepts risky shell operations, warns you of the potential danger, and requires interactive confirmation before executing any potentially harmful action.

Yes, this destructive command guardrail supports logging for compliance and audit purposes. It records risky actions and intercepted commands into an audit-ready log, allowing you to review incidents later and ensure compliance across your shared workspaces and production environments.

To override a safety warning when you intentionally need to run a risky shell command, you simply interact with the interactive warning prompt. The guardrail allows you to review the detected risk and explicitly choose to proceed with execution or cancel the operation.

You should not use a prompt-based guardrail for shell sessions in fully automated, non-interactive workflows where pausing for user confirmation would halt execution. Since the mechanism relies on interactive warnings to override destructive patterns, unattended automation pipelines would stall waiting for input.