kenryu42/cc-safety-net

A pre-execution guard for AI coding agents. It blocks destructive Git and file system commands, plus common attempts to access sensitive files, before a tool call runs. Supports Amp Code, Antigravity CLI, Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot CLI, Grok Build, Hermes Agent, Kimi Code, OpenClaw, OpenCode, and Pi.

What it solves

CC Safety Net prevents AI coding agents from executing destructive shell commands or accessing sensitive files (like SSH keys and .env files) before the tool call is executed. It acts as a security layer that blocks high-risk actions—such as rm -rf or git push --force—to prevent accidental or intentional data loss and secret leakage when using AI-powered CLIs.

How it works

The tool parses and analyzes commands before they run, ensuring that wrapping a command or reordering flags cannot bypass the protection. It integrates as a hook into various coding CLIs (e.g., Claude Code, Cursor, GitHub Copilot CLI) and can also be used as a Node.js library. Users can manage security policies via a GUI, install specialized rulebooks for cloud providers (AWS, Azure, GCP), and share these policies across teams via Git.

Who it’s for

Developers using AI coding agents and CLIs who want to ensure their local environment and sensitive credentials remain protected from autonomous agent actions.

Highlights

  • Multi-CLI Support: Works with a wide range of agents including Claude Code, Cursor, Gemini CLI, and GitHub Copilot CLI.
  • Customizable Safety Presets: Offers Standard, Strict, and Paranoid levels of protection.
  • Rulebooks: Extensible protection via official packs for Terraform, AWS, gcloud, and Azure.
  • Audit Trail: Maintains local logs of command decisions for review.
  • Library API: Can be embedded directly into other tools via an npm package.
  • Policy Sharing: Allows security policies to be committed to Git for team-wide consistency.

Related

  • Project
    Dicklesworthstone/destructive_command_guard

    A high-performance safety hook for AI coding agents that intercepts and blocks destructive commands like accidental deletions or database drops before they execute.

  • Project
    FailproofAI/failproofai

    Failproof AI is a Node.js/TypeScript library that hooks into 12 popular LLM‑coding and chat agents, records every model and tool call, and applies 39 built‑in safety policies (e.g., block API‑key leaks, sudo, rm‑rf, destructive SQL). Users can add custom policies in JavaScript, view full execution traces on a local dashboard (localhost:8020), run offline audits, and optionally use a hosted/self‑hosted service for fleet‑wide observability and alerts.

  • Project
    microsoft/agent-governance-toolkit

    A governance framework for autonomous AI agents that provides deterministic policy enforcement, identity management, and audit logging to prevent unauthorized actions.

  • Project
    jo-inc/camofox-browser

    An anti-detection browser server for AI agents that uses a C++ patched Firefox fork to bypass bot detection and provide token-efficient web interaction.

  • Project
    cosai-oasis/project-codeguard

    A security skills framework and ruleset that embeds secure-by-default practices into AI coding agents to prevent vulnerabilities during code generation and review.