kenryu42-claude-code-safety-net

Overview

Claude Code Safety Net is a plugin that acts as a protective layer for Claude Code, catching destructive git and filesystem commands before they execute. It monitors commands in real time and blocks potentially harmful operations such as force pushes, hard resets, and mass file deletions. Compatible with Claude Code, OpenCode, Gemini CLI, and GitHub Copilot CLI, it provides an essential safeguard for developers working with AI coding assistants.

Key Features

• Destructive command interception -- catches dangerous git commands (force push, hard reset, branch deletion) and filesystem operations before execution.
• Multi-tool compatibility -- works with Claude Code, OpenCode, Gemini CLI, and GitHub Copilot CLI, providing consistent protection across different AI coding tools.
• Beyond permission deny rules -- provides smarter protection than simple deny rules by analyzing command context and intent rather than just pattern matching.
• Sandbox-complementary -- works alongside sandboxing mechanisms to provide defense in depth, catching commands that sandboxing alone might not prevent.
• CI-tested with code coverage -- includes comprehensive CI testing and code coverage tracking via Codecov, ensuring reliability of the safety mechanisms.

Who is this for?

This plugin is essential for developers who use AI coding assistants and want to prevent accidental destructive operations on their repositories. It is especially valuable for teams where multiple developers use Claude Code or similar tools, providing a safety net that prevents costly mistakes like accidental force pushes or file deletions.