Run Claude Code in a Docker Container

Running Claude Code in a container gives you a sandboxed environment where it can’t touch your host filesystem, install random packages globally, or break your local setup.

Quick Start

docker run -it --rm \
  -e ANTHROPIC_API_KEY=$ANTHROPIC_API_KEY \
  -v $(pwd):/workspace \
  -w /workspace \
  node:20 bash -c "npm install -g @anthropic-ai/claude-code && claude"

Dockerfile for a Reusable Image

FROM node:20-slim

RUN npm install -g @anthropic-ai/claude-code
RUN apt-get update && apt-get install -y git jq curl && rm -rf /var/lib/apt/lists/*

WORKDIR /workspace
ENTRYPOINT ["claude"]

Build and run:

docker build -t claude-code .
docker run -it --rm \
  -e ANTHROPIC_API_KEY=$ANTHROPIC_API_KEY \
  -v $(pwd):/workspace \
  claude-code

When to Use Containers

• Untrusted repos: reviewing code you didn’t write
• CI/CD pipelines: reproducible environments across runs
• Team standardization: everyone uses the same toolchain
• Risky operations: let Claude run destructive commands safely

Mount Your Config

Pass your CLAUDE.md and settings into the container:

docker run -it --rm \
  -e ANTHROPIC_API_KEY=$ANTHROPIC_API_KEY \
  -v $(pwd):/workspace \
  -v ~/.claude:/root/.claude:ro \
  claude-code

The :ro flag mounts your config as read-only so Claude can’t modify your global settings.