Claude Code Container

Claude Code Container

A Docker container for running Claude Code in "dangerously skip permissions" mode.

https://github.com/user-attachments/assets/81c731d9-caeb-48cf-aa3e-65a48c55519e

Build the docker container and execute run_claude.sh to run an isolated version of claude code with access to the current working dir (readOnly:/workspace/input).

Variants

1. claude-standalone
Basic Claude Code container without any MCP servers configured. Clean, simple setup.

2. claude-with-mcp-example
Claude Code container with MCP servers pre-configured (e.g., Chonky Security Tools). Shows how to add MCP servers, configure them, and auto-trust their execution.

Quick Start

Prerequisites

1. Claude Code License: Ensure you have a valid Claude Code license
2. OAuth Token: Set your Claude Code OAuth token
3. Docker: Docker must be installed and running

Build and Run

Environment Variables

- CLAUDE_CODE_OAUTH_TOKEN: Your Claude Code OAuth token (required)

Run claude setup-token, login, save the resulting sk-* token.

Security Features

Container Security
- Non-root execution: Runs as user claude (UID 1001)
- Capability dropping: Minimal Linux capabilities
- Process limits: Resource constraints for safety (max 100 PIDs)
- Tmpfs mounts: Isolated temporary storage for /tmp and /workspace/temp
- Network isolation: Bridge network with no host access
- Security options: No new privileges allowed

Jailfree Mode
- Dangerous executions allowed: Pre-configured for full automation
- Auto-trusted workspace: No trust prompts during analysis
- Comprehensive tool permissions: Access to all tools via wildcard allowlist

MCP Server Integration (claude-with-mcp-example)

The MCP example shows how to integrate Model Context Protocol servers:

Adding Your Own MCP Server

1. Copy MCP to build…