careful

To prevent destructive bash commands in production, you need safety guardrails that intercept risky shell operations before execution. This skill implements a pre-tool use hook that analyzes shell commands for risky patterns like rm -rf and prompts for an explicit override before proceeding.

Yes, you can get warnings before running kubectl delete or git reset --hard in shared environments. The guardrail mechanism analyzes cluster management and git operations for destructive patterns, presenting an explicit override option to proceed.

A PreToolUse hook for blocking destructive database commands works by analyzing commands like DROP TABLE before execution. It identifies risky patterns in database operations and forces an explicit user override to proceed, preventing data loss.

The best way to add safety guardrails to AI-assisted deployment workflows is integrating a PreToolUse hook. This enforces safety by analyzing shell and deployment commands for destructive patterns and requiring explicit overrides before execution in production.

Yes, you can override the warning and still execute a force-push if needed. The guardrail presents an explicit override option after issuing the warning, allowing you to proceed with destructive commands intentionally when necessary.