Which local files are exposed to the container?

The architecture follows a principle of least privilege, mounting only essential files like the proxy CA certificate and basic configuration while keeping secrets and databases on the host.

What should I do if the container asks for onboarding?

This indicates the claude.json mount is missing; the skill provides the specific JSON structure needed to mark onboarding as complete automatically.

Can I use this for automated CI/CD workflows?

Yes, the skill details a specific 'Print Mode' designed for non-interactive environments where Claude can output responses directly to logs and exit.

How does this skill handle sensitive API keys?

It utilizes a host-side proxy to inject real OAuth tokens into outgoing HTTPS requests, ensuring the container itself only interacts with placeholder credentials.

Claude Code Container Proxy

Name: Claude Code Container Proxy
Author: shepherdjerred

byshepherdjerred

•

デプロイメントとDevOps

Enables secure, zero-credential execution of Claude Code within Docker containers using a TLS-intercepting host proxy.

This skill provides a comprehensive architectural framework for running Claude Code in isolated Docker environments without exposing sensitive API keys or OAuth tokens. By leveraging the clauderon proxy, it intercepts HTTPS requests at the host level to inject authentication headers, ensuring that containers operate in a zero-trust state. This is ideal for developers who need to run Claude Code in CI/CD pipelines, test dangerous commands in sandbox environments, or debug complex network and authentication issues without compromising their primary credentials.

主な機能

01Dual execution modes supporting both interactive sessions and non-interactive print modes

02Comprehensive audit logging for monitoring proxied Anthropic API requests

034 GitHub stars

04Host-side TLS interception for secure OAuth token injection

05Automated onboarding bypass via pre-configured mount patterns

06Zero-trust container architecture that hides real credentials from the environment

ユースケース

01Running Claude Code in CI/CD pipelines for automated code review and generation

02Testing and debugging containerized AI agent workflows without sharing global secrets

03Executing CLI commands in isolated Docker sandboxes to prevent local system interference

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add shepherdjerred/monorepo zero-credential-claude

For use in Claude.ai and ChatGPT

Download Skill