careful

Intercept destructive Bash commands and require user confirmation before execution.

Updated Apr 3, 2026
One-click install
npx skills add https://github.com/lyh2/gstack-deepseek --skill careful-lyh2
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: careful
Source: https://github.com/lyh2/gstack-deepseek/tree/main/careful
Command: npx skills add https://github.com/lyh2/gstack-deepseek --skill careful-lyh2

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Safety guardrails for destructive Bash commands. Warns before rm -rf, git reset --hard, kubectl delete, and similar destructive operations in prod or shared environments. Users can override each warning when appropriate.

Core Features & Use Cases

  • Detects dangerous commands across common tooling and surfaces a clear warning before execution.
  • Provides per-command override controls and safety checks to prevent accidental data loss.
  • Use cases include safeguarding production deployments, debugging in shared environments, and teaching teams safe command habits.

Quick Start

Run a destructive command in a guarded Bash session to see the warning and choose to proceed or cancel.

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 prevent accidental execution of destructive Bash commands?

Safety guardrails intercept destructive Bash commands and require user confirmation before execution. This mechanism analyzes input commands, emits a warning, and allows override only when permitted to prevent accidental data loss.

Can I get a confirmation prompt before running rm -rf or git reset --hard in a shared environment?

Yes, a confirmation prompt can be triggered before running rm -rf or git reset --hard in shared environments. The system detects dangerous commands across common tooling and surfaces a clear warning before execution proceeds.

How do I block kubectl delete operations in production environments?

To block kubectl delete operations in production, you can intercept these high-risk commands and require explicit user confirmation. The system targets destructive operations across prod environments to prevent accidental resource deletion.

Is there a way to override the safety warning if I intentionally need to execute a destructive command?

Yes, users can override the safety warning when appropriate. The system provides per-command override controls and safety checks, allowing you to proceed with intentional destructive operations after acknowledging the prompt.

Does this safety mechanism work across different command-line tools like git and kubectl?

Yes, the safety mechanism works across different command-line tools like git and kubectl. It detects dangerous commands across common tooling, including rm, git reset, and kubectl, surfacing warnings before execution.

What are the limitations of using command interception for preventing data loss?

The limitation of command interception is that it relies on user judgment, as safety warnings can be overridden. It serves as a guardrail against accidental data loss but cannot block intentional execution if the user permits the override.