careful

Intercept destructive Bash commands and require confirmation before execution.

Updated Apr 11, 2026
One-click install
npx skills add https://github.com/Scivor/helm --skill careful-scivor
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: careful
Source: https://github.com/Scivor/helm/tree/main/skills/gstack/careful
Command: npx skills add https://github.com/Scivor/helm --skill careful-scivor

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Destructive commands can cause data loss or service disruption. This Skill intercepts risky Bash invocations, warns the user, and requires explicit confirmation before proceeding (with an easy override when safety is desired).

Core Features & Use Cases

  • PreToolUse hook that runs a check-careful.sh script prior to tool execution to detect dangerous patterns.
  • Guardrails for common destructive actions (rm -rf, git reset --hard, kubectl delete, etc.) with a clear warning message and an option to proceed.
  • Designed for development, staging, and shared environments where accidental destructive commands could impact others.

Quick Start

Enable careful mode in a session and run a potentially destructive Bash command to see the warning prompt.

Frequently Asked Questions about careful

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I prevent accidental destructive bash commands like rm -rf in development environments?

To prevent accidental destructive bash commands, you need safety guardrails that intercept risky invocations and require explicit confirmation before proceeding. This Skill uses a PreToolUse hook to detect patterns like rm -rf and prompt the user.

Can I override the warning prompt for destructive commands when I am certain they are safe?

Yes, you can override the warning prompt for destructive commands. The Skill requires explicit user confirmation before proceeding, allowing an easy override when safety is desired and the destructive action is intentional.

Does this safety guardrail intercept git reset --hard and kubectl delete operations?

Yes, the safety guardrail intercepts git reset --hard and kubectl delete operations. It implements a check script that detects these common destructive actions, returning a permission decision prompt to warn the user before execution.

What is the best way to add warning guardrails for shared staging environments?

The best way to add warning guardrails for shared staging environments is a PreToolUse hook that intercepts risky bash invocations. This prevents accidental service disruption by requiring explicit confirmation for destructive actions.

How does the PreToolUse hook detect destructive commands before execution?

The PreToolUse hook detects destructive commands by executing a check-careful.sh script prior to tool execution. This script identifies dangerous patterns and returns a permission decision prompt when risk is detected, logging events locally.

Are there limitations to using bash safety guardrails in automated workflows?

A limitation of bash safety guardrails in automated workflows is the required explicit confirmation. Since the system warns before destructive actions and requires user input, automated pipelines may stall unless an override is configured.