Claude Code on a VPS: run it on a Linux server and drive it from your phone

What you need

• A Linux server you can SSH into, from a provider such as DigitalOcean or OVHcloud. Any distribution works for the Claude Code CLI; Claude Desktop needs Ubuntu 22.04 or later, or Debian 12 or later.
• A Claude Pro, Max, Ultra, Team or Enterprise plan. Remote Control doesn't work with API keys, and on Team or Enterprise an Owner turns it on first.
• The Claude app on your phone (iOS or Android), or claude.ai/code in a browser.
• Optionally, a desktop environment and a remote desktop on the server, for the times Claude needs a website you have to sign in to.

How to set up Claude Code on a Linux VPS

1. Rent a server and connect to it

   Create a Linux server with a provider such as DigitalOcean or OVHcloud and connect to it with SSH. Choosing Ubuntu 22.04+ or Debian 12+ keeps Claude Desktop as an option; on any other distribution you use the CLI.

2. Install Claude Code and sign in

   Install the Claude Code CLI with Anthropic's install script. It runs in a terminal (Anthropic lists Ubuntu 20.04+ and Debian 10+ among the supported systems), so it suits a headless server you only reach over SSH. On Ubuntu 22.04+ or Debian 12+ you can install Claude Desktop instead, with apt; it's in beta on Linux (see Claude Desktop on Linux). Then run claude and sign in with your Claude account, not an API key.

   curl -fsSL https://claude.ai/install.sh | bash

3. Start Remote Control in server mode

   From your project folder, run claude remote-control. It starts a server that can host several Claude Code sessions at once, and you can open each of them from your phone. Add --chrome to turn on Chrome integration in those sessions. To have every interactive session connect instead, set remoteControlAtStartup to true in ~/.claude/settings.json.

   # Start a tmux session, so the server keeps running after you close SSH
   tmux new -s claude
   
   # Inside tmux, from your project folder: run the Remote Control server and
   # restart it whenever it exits (it gives up after ~10 minutes offline)
   cd ~/my-project
   while true; do claude remote-control; sleep 10; done
   
   # Detach with Ctrl+B, then D. Reattach later with:
   tmux attach -t claude

4. Keep it running after you log out

   Remote Control only works while the claude process runs: close the terminal and its sessions go offline. Run it inside tmux or screen so it survives the SSH connection closing, or as a service. Server mode also exits after about 10 minutes without a network, so restart it automatically, as the loop above does. For permission modes and usage limits on long runs, see how to run Claude Code 24/7.

5. Open it from your phone

   Open the Claude app on iOS or Android, or claude.ai/code, and pick a session. Claude runs on the server; the phone shows the conversation and sends your messages. Setup details and fixes are in the Remote Control guide. To message Claude Code from Telegram instead, see Claude Code channels.

Source: Anthropic's Remote Control documentation

Add a desktop for signing in to websites

A server you reach over SSH has no screen, which is fine until Claude needs a site you have to sign in to. Add a remote desktop and you can do that sign-in yourself, then let Claude use the browser.

• Install a desktop environment and a remote desktop server on the VPS, and connect from your laptop when you need to sign in somewhere.
• Install Google Chrome or Microsoft Edge with the Claude in Chrome extension, and sign in to the sites Claude should use. Claude in Chrome works on Linux; it isn't supported in WSL.
• Start the server with claude remote-control --chrome so its sessions can use that browser.
• Reach the remote desktop over Tailscale, which puts your devices on a private network, a tailnet, where only they can reach each other. See what Tailscale is.
• With a desktop on Ubuntu or Debian, Claude Desktop is an option as well as the CLI.

What a Linux server can't do

If any of these matter, use a Mac mini at home instead, the setup our founder uses.

• No computer use. Claude can't operate apps and the screen on Linux: in the CLI computer use is macOS only, and in Claude Desktop it's a research preview on macOS and Windows.
• No iOS development. Xcode and the iOS Simulator run only on macOS, so a Linux server can't build or run an iPhone app.
• A smaller desktop app. Claude Desktop on Linux is a beta for Ubuntu and Debian, without computer use, Cowork or dictation.
• A process that has to stay up. As with any Remote Control setup, the sessions are reachable only while claude runs on the server.

Mac mini vs Linux VPS for Claude Code

Give the server a standing job

A server that never sleeps answers whenever you message it, and waits when you don't. Crewmark turns a Claude Code conversation on it into an AI employee with a role, goals, working hours and a heartbeat: during working hours the crewmark CLI, which runs on Linux as well as macOS and Windows, wakes the employee to work through its issues on a shared board. The computer it runs on has to stay on, and a VPS does.

1. Hire an employee

   Pick a role from the AI employee roles, such as software engineer or SEO specialist, and write its goals and KPIs.

2. Connect Claude Code on the server

   Paste the setup link into a Claude Code conversation on the VPS. It installs the crewmark CLI and links that conversation to the employee. Codex works too.

3. Set working hours and a heartbeat

   Weekdays only or every day, hours that can run past midnight, and a wake-up every 15, 30, 60, 120 or 240 minutes.

4. Assign issues from your phone

   Create and assign issues on the board at crewmark.ai from any browser; assigning one wakes the employee right away. When it needs a review or a decision it sets the issue to Needs action and asks you there. Jobs to give it are in Claude Code use cases.

FAQ