careful

Detect dangerous Bash command patterns and pause execution with a permission prompt.

Updated Apr 13, 2026
One-click install
npx skills add https://github.com/mgher668/surf-ai --skill careful-mgher668
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: careful
Source: https://github.com/mgher668/surf-ai/tree/main/.agents/skills/gstack/careful
Command: npx skills add https://github.com/mgher668/surf-ai --skill careful-mgher668

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Safety mode helps prevent destructive shell commands from executing by warning users before proceeding, reducing risk in production and live environments.

Core Features & Use Cases

  • Destructive command detection for common risky operations (rm -rf, git reset --hard, kubectl delete, Docker prune, and more).
  • Safe exceptions and allowlists for certain targets to avoid blocking legitimate maintenance tasks.
  • PreToolUse integration to intercept commands and present a warning with a choice to continue or cancel.
  • Local analytics logging of skill usage to help teams audit risky actions.

Quick Start

Enable careful mode and run Bash commands; you will receive warnings before any destructive operations.

Frequently Asked Questions about careful

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

FAQPage Schema
How do I guard destructive bash commands with a safety prompt?

You can guard destructive bash commands by enabling a safety mode that detects risky patterns like rm -rf or git reset --hard and pauses execution with a permission prompt. This prevents accidental data loss in live environments.

What destructive commands does a bash safety guard detect?

A bash safety guard detects common destructive operations such as rm -rf, git reset --hard, kubectl delete, and docker prune. It analyzes these input commands and intercepts execution to present a warning before proceeding.

Can I use safety guardrails for kubectl delete and docker prune in production?

Yes, safety guardrails apply to development, staging, and production workflows. They intercept commands like kubectl delete or docker prune and require explicit permission before executing to prevent outages and data loss.

How do I allowlist certain targets to avoid blocking legitimate maintenance tasks?

You can configure safe exceptions and allowlists for certain targets to avoid blocking legitimate maintenance tasks. This allows specific destructive commands to execute without triggering the safety warning.

Does a bash command audit log track when destructive commands are intercepted?

Yes, safety guardrails log the event locally for audit when a dangerous command pattern is detected. This local analytics logging helps teams audit risky actions and review when warnings were triggered.

Why do I need a confirmation prompt for git reset --hard?

A confirmation prompt for git reset --hard is needed because it is a destructive command that can cause irreversible data loss. The prompt intercepts execution to ensure the operation is intentional and authorized.