How to set up a spare Mac for Claude Code to control remotely

Key Takeaways

• Running Claude Code with full permissions on a dedicated spare Mac isolates security risks from your main machine
• The setup enables remote control via SSH from any device, including the Claude mobile app
• Unlike containers, a spare Mac provides access to GUI applications like Unity that Claude can control visually

Developer YK Dojo has published a detailed guide for turning a spare Mac into a dedicated machine that Claude Code can control autonomously. The setup addresses a real problem: Anthropic's Claude Code agent is powerful, but running it with the --dangerously-skip-permissions flag on your primary computer carries risk. A separate machine with nothing important to lose changes that equation.

Why a spare Mac instead of a container?

YK Dojo built a containerized environment for Claude Code previously, but found limitations that pushed him toward dedicated hardware. Containers still run on your main machine, meaning network traffic and system resources remain shared. More importantly, containers cannot run macOS-only applications. If you want Claude to operate Unity for game development, or any other GUI app through Anthropic's computer use feature (clicking, dragging, typing), a container won't cut it.

The spare Mac approach also lets you talk to Claude Code from the official Claude mobile app. YK notes he often prefers Claude Code over standard Claude on mobile because it's "often more capable." The setup works with a Claude Pro subscription ($20/month), which some founders already have.

The core setup steps

The guide recommends starting fresh. If your spare Mac has any data you don't want the agent accessing, wipe it first. Claude Code with full permissions can reach anything on the machine.

After that, the configuration involves four main steps:

1. Enable SSH on the target Mac (sudo systemsetup -setremotelogin on)
2. Grant passwordless sudo so the agent can run admin tasks without prompts
3. Set up hostname resolution so your main machine can reliably reach the spare
4. Install your SSH key for password-free connections

The passwordless sudo step is where most of the security tradeoff happens. You're creating a rule that lets the user account run any command without authentication. On a production machine, this would be reckless. On a wiped spare Mac with nothing sensitive, it's the whole point.

Using hostname over IP address

YK recommends connecting via hostname (e.g., macbook-pro.local) rather than IP address. The IP can change after reboots or DHCP lease renewals. The .local hostname stays constant. One gotcha: each Mac on your network needs a unique hostname. If two machines share a name, your connections might route to the wrong one.

To check your spare Mac's hostname, run scutil --get LocalHostName. To change it: sudo scutil --set LocalHostName newmacbook, which makes the address newmacbook.local.

What tasks actually benefit from this setup?

YK mentions two categories: research tasks and development tasks he doesn't want running on his primary machine. The computer use capability is the key differentiator here. Claude Code can now see the screen, move the cursor, click buttons, and type text. That opens up workflows impossible in a terminal-only environment.

For startup founders, think: automated testing that requires visual verification, prototype building in GUI tools, or any repetitive computer work that's hard to script but easy to demonstrate. If you can show Claude what to do by example, computer use lets it repeat the process.

Security considerations

Let's be clear about what you're enabling. Passwordless sudo plus the --dangerously-skip-permissions flag means Claude can do anything on that machine without asking. It can install software, modify system files, send network requests, and access any data on the disk. Anthropic named that flag "dangerously" for a reason.

The mitigation strategy is isolation, not restriction. You accept that the spare Mac is fully compromised from a security perspective. You just ensure there's nothing on it worth protecting and that it can't reach sensitive resources on your network.

Remote access from anywhere

Once configured, you can SSH into the spare Mac from your main machine and run Claude Code commands directly. The setup also works with the Claude mobile app, letting you issue tasks from your phone while away from your desk. YK specifically calls out this mobile access as one of the main benefits. Starting a long-running task before a meeting, then checking results after, becomes practical.

Frequently Asked Questions

Does this setup require a Claude Pro subscription?

Claude Code works with Claude Pro ($20/month). You can use your existing subscription's usage allocation for tasks on the spare Mac.

Can I use a Mac mini instead of a MacBook?

Yes. The guide applies to any Mac. A Mac mini is actually ideal since it's designed for always-on operation and doesn't tie up a portable machine.

What happens if Claude Code runs a destructive command?

On this setup, it executes. That's the point of passwordless sudo. The safety model is isolation: nothing on the spare Mac should matter if lost or corrupted.

Can Claude Code control apps like Figma or Slack through computer use?

In theory, yes. Computer use lets Claude see the screen and interact with GUI elements. App support depends on whether the app responds to standard macOS input methods.

Is this setup reversible?

Yes. To disable passwordless sudo, delete the /etc/sudoers.d/<user>-nopasswd file. To disable SSH access, run sudo systemsetup -setremotelogin off.

Source: Hacker News: Best