How to run Claude Code on more than one machine.

What Claude Code is, briefly

Claude Code is Anthropic's agentic coding tool. It reads your codebase, edits files, runs commands and works with tools like git. It runs in a terminal, in IDE extensions, in a desktop app and in the browser. Most of those surfaces need a Claude subscription or an Anthropic Console account. For the current details, Anthropic's documentation is the source of truth.

The key fact for this guide: when you run it in a terminal, Claude Code works on the machine and in the folder where you started it. To work on a second machine, you need it on that machine too.

Way 1: Install and sign in on every machine

The direct route. On each computer, install Claude Code and start it inside the project. Anthropic's recommended native installer at the time of writing:

# macOS, Linux, WSL
curl -fsSL https://claude.ai/install.sh | bash

# Windows PowerShell
irm https://claude.ai/install.ps1 | iex

Then, in the project folder:

cd your-project
claude

You'll be asked to log in on first use. If the ANTHROPIC_API_KEY environment variable is set, Claude Code asks you to approve that key instead.

Where it hurts: every machine is its own install, its own sign-in and its own set of updates. On a headless server, the sign-in step is the awkward one, because the browser you'd normally use is on a different computer.

Way 2: SSH in and keep sessions alive with tmux

For Linux machines, most people SSH in and run Claude Code over the connection. The catch is that closing your laptop or losing Wi-Fi ends the SSH session, and the agent with it. A terminal multiplexer like tmux fixes that:

ssh build-box
tmux new -s api      # start a named session
cd ~/code/api && claude
# detach with Ctrl-b then d; reattach later with:
tmux attach -t api

Where it hurts: you now have one terminal per machine, and you're the one switching between them. Nothing collects the results. Nothing tells you when one finishes.

Running two sessions on the same machine

Sometimes the second "machine" you need is really a second copy of the repo. Two agents editing the same working folder will trip over each other. Git worktrees give each one its own checkout of the same repository:

git worktree add ../api-fix-tests -b fix-tests
git worktree add ../api-docs -b update-docs
# run one Claude Code session in each folder

Each session gets a clean folder and its own branch, and you merge the results like any other branches.

Way 3: Let Anthropic's cloud do the running

Anthropic also offers Claude Code on the web, which runs tasks in your browser with no local setup and can run several tasks in parallel. Its Remote Control feature lets you continue a local session from your phone or another browser.

Where it hurts: cloud sessions run on Anthropic's infrastructure. They don't reach the local database, the files on your NAS or the GPU under your desk. Remote Control helps you check on a session, but each session still lives on one machine.

Make every machine's agent follow the same rules

Whichever way you go, commit a CLAUDE.md file to the root of the repository. Claude Code reads it at the start of every session, so coding standards, architecture notes and review checklists travel with the code to every machine that checks it out. If your repo already has an AGENTS.md for other coding agents, Claude Code can read that too.

Where the manual approach breaks down

Two machines are manageable by hand. Past that, the overhead compounds:

• Installs, sign-ins and updates multiply by the number of machines.
• Context is stuck in whichever terminal you used. The session on the server doesn't know what the session on the desktop just did.
• There's no single report. You read each terminal and stitch it together yourself.
• Permissions are all-or-nothing per machine unless you configure each one carefully.

How Syrebrum handles it

Syrebrum treats coding agents as workers on your own machines. From the command center on your Windows PC you can:

• Have Sy install Claude Code or OpenCode on any connected machine and walk you through signing in, with the sign-in link shown on your main PC.
• Hand work to Claude Code, OpenCode or GitHub Copilot's agent on whichever machine has the code.
• Set a permission level for each folder, so an agent works only where you've allowed it.
• Run agents on different machines at the same time and read the results in one place.
• Keep a shared memory of machines, projects and what has been done, with secrets masked.

See how coding agents work in Syrebrum.

Questions