careful

Warn before executing destructive Bash commands via a PreToolUse hook.

1|Updated May 6, 2026
One-click install
npx skills add https://github.com/yckkkk/reo --skill careful-yckkkk
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: careful
Source: https://github.com/yckkkk/reo/tree/main/.agents/skills/careful
Command: npx skills add https://github.com/yckkkk/reo --skill careful-yckkkk

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Safety guardrails for destructive shell commands, warning before executing dangerous operations such as rm -rf, git reset --hard, and kubectl delete, to prevent accidental data loss.

Core Features & Use Cases

  • Intercepts Bash commands and flags potentially destructive actions.
  • Allows user override when permissions are explicitly granted, with auditing.
  • Logs usage to ~/.gstack/analytics/skill-usage.jsonl for monitoring.

Quick Start

Enable careful mode by running a command with destructive intent and observe the warning prompt before proceeding.

Frequently Asked Questions about careful

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

FAQPage Schema
How do I add safety guardrails to warn before executing destructive bash commands?

Bash safety guardrails intercept destructive shell operations like rm -rf and prompt for user confirmation before execution, preventing accidental data loss across development and production environments.

What destructive shell operations should trigger a warning prompt before execution?

Destructive operations include file deletions, database drops, and repository history rewrites such as git reset --hard, kubectl delete, and rm -rf, all of which risk irreversible data loss.

Can I override the warning and still execute a destructive command if I explicitly grant permission?

Yes, you can override the guardrail warning and proceed with destructive commands when permissions are explicitly granted, and the override action is recorded for auditing purposes.

Does this safety guardrail work for bash workflows across staging and production environments?

Yes, the guardrail is applicable to Bash-powered workflows across development, staging, and production environments, intercepting dangerous shell operations consistently throughout the deployment pipeline.

How is audit logging handled when a destructive command is intercepted or overridden?

Audit logging records usage details to ~/.gstack/analytics/skill-usage.jsonl, capturing both intercepted warnings and explicitly overridden destructive commands for monitoring and compliance tracking.

What are the limitations of using a PreToolUse hook to detect destructive bash patterns?

The PreToolUse hook relies on pattern detection for known destructive commands, meaning highly unusual or obfuscated shell operations might bypass the guardrails if they do not match the recognized dangerous patterns.