guard

Enforce a directory edit boundary and warn before destructive commands.

Updated Mar 22, 2026
One-click install
npx skills add https://github.com/jeisenback/original-curse-killer --skill guard-jeisenback
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: guard
Source: https://github.com/jeisenback/original-curse-killer/tree/main/.claude/skills/guard
Command: npx skills add https://github.com/jeisenback/original-curse-killer --skill guard-jeisenback

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Protects critical projects from accidental destructive edits by enforcing an editable boundary and warning before dangerous commands.

Core Features & Use Cases

  • Combines destructive-command warnings with an enforceable edit boundary to prevent risky changes in prod or live systems.
  • Allows interactive boundary selection of a directory and stores the boundary state for future sessions.
  • Provides safety prompts and guards across development workflows, incident response, and routine maintenance.

Quick Start

Ask the user for a directory to protect, then enable guard mode by setting the freeze boundary and enabling destructive-command warnings.

Frequently Asked Questions about guard

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

FAQPage Schema
How do I prevent destructive commands from running in a production environment?

To prevent destructive commands in production, you can enforce an edit boundary that warns before risky changes and blocks accidental edits across interactive sessions and automation workflows.

What is an edit boundary and how does it protect live systems?

An edit boundary is a directory-wide protection limit that prevents destructive edits outside the specified scope, safeguarding live systems during development, incident response, and routine maintenance.

Can I set a directory freeze boundary that persists across multiple sessions?

Yes, you can select a directory to protect and the freeze boundary state is stored in a state file, ensuring the edit boundary persists for future interactive and automated sessions.

Does guard mode work with automation workflows or only interactive sessions?

Guard mode works across both interactive sessions and automation workflows, integrating with helper hooks to warn on destructive commands during development and operations on live systems.

What is the best way to add safety prompts for risky shell commands during maintenance?

The best way to add safety prompts is enabling guard mode, which integrates with helper hooks to warn before dangerous commands and enforces a directory-wide edit boundary during maintenance.

When should I not use a directory-wide edit boundary for command safety?

You should avoid using a directory-wide edit boundary when your workflow requires frequent destructive edits inside the protected directory, as the freeze boundary will actively block those changes.