guard

To prevent destructive commands in production, you can enable guard mode to activate pre-use checks that warn before potentially dangerous operations execute. This enforces directory boundaries to restrict edits to approved safe paths during live service maintenance.

To restrict file edits to a specific directory boundary, specify the target directory path when enabling guard mode. This blocks any modifications outside the designated boundary, ensuring only approved changes within the safe directory are allowed.

Pre-tool hooks for safety checks intercept tool usage before execution to enforce directory boundaries and warn about destructive operations. They rely on sibling skills like /careful and /freeze to perform the actual safety validations.

Yes, you need the /careful and /freeze skills because guard relies on these sibling integrations to perform pre-tool safety checks. Guard references them to satisfy cross-skill dependencies and execute the actual validation logic before tool use.

Yes, you can use guard mode for live service maintenance without blocking safe edits because it enforces a directory boundary rather than a total lockdown. It prompts for a target directory and only blocks edits outside that approved path while warning about destructive commands.

You should not rely on guard mode for production protection if your workflow lacks sibling /careful and /freeze skills, as guard requires these cross-skill dependencies to execute its pre-use checks and cannot enforce directory boundaries independently without them.