Claude Code

Your own Claude Code, one tab away from your work

Pair the Claude Code on your machine with your workspace. Use it from a chat tab, let it work with your to-dos, and answer its permission prompts from your phone.

Follow the setup tutorial Get the companion app

How it works

It runs on your machine. Your workspace just connects to it.

A small companion app, @openwop/agent-host, pairs your machine with your workspace using a one-time code you type yourself. It connects out to the workspace and never opens a port on your machine.

Claude Code keeps running where your code lives, in the folder you choose, signed in the way you already sign it in. Your Claude credential never leaves your machine, and this app never sees it.

What you can do

Four ways to work with it

Start with the terminal tab; the rest is there when you need it.

Set up

Three steps

About ten minutes with Node.js 22 and Claude Code already installed.

Before you turn it on

Opt-in, and on your own credential

Claude Code is off until your workspace admin enables it. Signing Claude Code in with an Anthropic API key, Amazon Bedrock or Google Vertex is the supported path; using a personal Claude login is at your own risk, and the approval page shows the relevant terms before you pair.

You can revoke a device at any time from the Claude Code page, and it stops working on its next request.

Get started

Bring your Claude Code into your workspace

The tutorial walks you through pairing, the chat tab and inbox approvals.

Start the tutorial