Does it work on Windows systems?

Yes, Damage Control includes a dedicated pathway to add PowerShell and CMD patterns, ensuring security across both Unix and Windows environments.

Can I customize security rules for different projects?

Absolutely. Damage Control supports global, project-specific, and local settings files, allowing you to define different protection levels for different codebases.

Can I protect specific folders from being modified?

Yes, you can define zero-access, read-only, and no-delete paths to prevent the AI from viewing, modifying, or deleting sensitive directories.

How does Damage Control block commands?

It uses PreToolUse hooks to intercept Claude's tool calls and matches them against a configurable list of dangerous patterns defined in a shared YAML configuration.

What happens when a command is blocked?

The system returns a specific exit code that halts the execution and informs the user that the operation was blocked by security policies.

Damage Control Security Hooks

Name: Damage Control Security Hooks
Author: gravity9-tech

bygravity9-tech

0•

Security & Testing

Protects your workspace by intercepting and blocking dangerous shell commands and unauthorized file modifications.

Damage Control is a defense-in-depth security system for Claude Code designed to prevent accidental or malicious system damage during AI-assisted development. By utilizing PreToolUse hooks, it monitors tool requests to block destructive command patterns, enforces granular file access permissions—including zero-access and read-only paths—and triggers interactive user confirmations for high-risk operations. Whether working globally or within specific projects, this skill provides a robust safety net for developers using Claude Code to automate tasks safely.

Key Features

01Granular Path Protection (Zero Access, Read-Only, No-Delete)

02Multi-level configuration (Global, Project, and Local overrides)

030 GitHub stars

04Cross-platform support for Unix, PowerShell, and CMD environments

05Interactive 'Ask' triggers for risky-but-valid operations

06Command Pattern Blocking for destructive operations like 'rm -rf'

Use Cases

01Preventing accidental hard resets or destructive git operations

02Securing sensitive credential files and environment variables from AI access

03Enforcing read-only constraints on critical system configuration files

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add gravity9-tech/pandora-marketplace damage-control

For use in Claude.ai and ChatGPT

Download Skill