careful

To prevent destructive shell commands, you need safety guardrails that intercept execution and require explicit confirmation. This skill uses a PreToolUse hook in Bash environments to analyze commands for patterns like rm -rf, force-push, and kubectl delete, returning a permission decision before they run.

Yes, you can intercept and warn before running kubectl delete or force-push in a Bash shell. The skill implements real-time pattern matching for these specific destructive actions, surfacing a clear warning message and requiring confirmation before proceeding with execution.

Shell safety pattern matching works by analyzing tool commands for destructive patterns before execution. When a command matches known dangerous actions, the PreToolUse hook intercepts it and returns a permission decision, providing an opt-out override option if the operation is intentional.

This destructive command guardrail applies specifically in Bash shells and development environments. It relies on a PreToolUse hook mechanism to analyze commands, so its shell safety features are designed for that context rather than general system-wide command interception.

The best way to monitor unsafe operations without exfiltrating shell data is using lightweight local analytics. This skill logs guarded events locally to help teams track destructive command attempts and monitor unsafe operations without sending sensitive command data externally.

Yes, you can override the warning if a destructive command is intentional. The skill provides an opt-out override option alongside its real-time pattern matching, allowing you to proceed with guarded actions like git reset --hard after acknowledging the warning.