dangerously

Run Claude Code autonomously in an isolated Docker container.

Claude Code's --dangerously-skip-permissions flag lets agents act without interruption — but Anthropic warns you to only use it inside a sandboxed environment. dangerously spins up that sandbox for you.

Requirements

• Docker
• Claude Code installed and authenticated on your machine
  • npm install -g @anthropic-ai/claude-code
  • Run claude once and complete login before using this tool

Install

``` npm install -g dangerously ```

Usage

Navigate to your project folder and run:

``` dangerously ```

Claude Code launches inside an isolated Docker container. File system changes are restricted to your current directory. Your host machine stays safe.

How it works

• Runs Claude Code in a fresh Docker container per session
• Mounts your current directory as the working directory
• Passes your Claude credentials from ~/.claude.json into the container
• Runs as a non-root user with --dangerously-skip-permissions enabled