careful

To prevent accidental data loss from destructive bash commands, a pre-run hook analyzes risky patterns like rm -rf and returns an 'ask' warning before execution. This intercepts potentially harmful operations in production environments and shared workspaces.

When debugging live systems, you should guard against rm -rf, git reset --hard, kubectl delete, DROP TABLE, and force-push. A pre-run hook detects these destructive operations and prompts an explicit warning before allowing execution.

Yes, you can override safety warnings for specific destructive operations on a per-command basis. The guardrail intercepts risky patterns and asks for permission, allowing you to explicitly approve an override when you intend to execute the destructive action.

A prehook detects risky shell commands by analyzing the command string before execution and matching it against known destructive patterns. When it identifies operations like git reset --hard or kubectl delete, it returns a permission decision to ask before proceeding.

Safety guardrails for local development rely on pattern matching, so limitations include potential false positives or missing novel destructive commands. They act as a proactive warning layer, but you still maintain final responsibility when overriding the prompt.

The best way to add safety mode when touching prod databases is to enable a pre-run hook that intercepts SQL DROP and TRUNCATE statements. It logs usage analytics and returns an 'ask' warning before any potentially destructive database action executes.