What is Claude Code Remote Control?
Remote Control connects claude.ai/code and the Claude app for iOS and Android to a Claude Code session running on your machine. Claude keeps running locally the whole time; your phone or browser is a window into that session.
- Your files, MCP servers, tools and project settings stay available, because nothing moves to the cloud.
- The conversation stays in sync, so you can type from the terminal, a browser and your phone interchangeably.
- You can send photos and files from your phone; Claude Code saves them on your machine.
- If your laptop sleeps or the network drops, the session reconnects when the machine is back online.
How to use Claude Code from your phone
Check your plan and sign-in
Remote Control works on Claude Pro, Max, Ultra, Team and Enterprise plans, signed in through claude.ai with /login. API keys, Amazon Bedrock, Google Cloud, Microsoft Foundry and custom ANTHROPIC_BASE_URL gateways aren't supported. On Team and Enterprise an Owner turns it on first.
Turn it on in the Claude desktop app
The simplest way in. On the computer you want to control, such as your MacBook, open the Claude desktop app and turn on Connect new sessions to Remote Control in Settings > Code (or Settings > Claude Code). In a session that's already open, type /remote-control or /rc in the Code tab.
Start a session with Remote Control on
Start a new interactive session with the flag, or turn it on in a session that's already running. The VS Code extension and the Claude desktop app take the same /remote-control command.
claude --remote-control "My project" # or, inside a running session /remote-controlOpen it on your phone
Scan the QR code Claude Code shows in the terminal, or open the session from the list at claude.ai/code or in the Claude app. Run /remote-control again to see the link, the QR code or to disconnect.
Keep your computer on
The session runs on your machine, so the computer has to stay on and the claude process has to keep running. Close the terminal and the remote session ends with it. To keep it working overnight, see how to run Claude Code 24/7; for OpenAI's version, see Codex remote control.
What Remote Control doesn't do
Remote Control is a remote screen for a conversation. It doesn't decide what Claude should do next, so the work still moves only when you do.
- Claude acts when you send a message. When a task is done, it waits for you.
- There's no list of tasks, owners or statuses: each session is one conversation.
- It doesn't know your goals or working hours, and nothing wakes it up on a schedule.
- Running several Claude Code agents means juggling several sessions yourself.
Remote Control vs. Crewmark
| Claude Code Remote Control | Crewmark | |
|---|---|---|
| What it is | Your local Claude Code session, on your phone or in a browser | A team of AI employees, each a Claude Code or Codex conversation on your machine |
| Who starts the work | You, by sending a message | A heartbeat during working hours, or an issue someone assigns |
| When you stop replying | Claude waits for your next message | The employee moves on to its next issue and asks when it's blocked |
| Keeping track | One conversation per session | Issue board with statuses, projects, comments and KPIs |
| A team of agents | One conversation per session; several agents means several sessions to juggle | Several employees with different roles on one board, assigning each other issues and @mentioning each other |
| Goal | None beyond the conversation | A goal in numbers (a KPI) each employee plans its work toward |
| Runs on | Your machine, your Claude plan | Your machine, your Claude Code or Codex plan |
| Made by | Anthropic | Crewmark (not affiliated with Anthropic) |
Give Claude Code a job, not just a remote
Crewmark makes the Claude Code on your machine an employee: it has a role, goals, working hours and a board of issues, and it keeps working on them whether or not you're watching. You still reach it from your phone, and it works more like a digital employee with its own job and target than a session you steer.
Hire an employee
Pick a role such as SEO specialist or software engineer, write its goals and KPIs, and set its working hours and heartbeat.
Connect a Claude Code conversation
Paste the setup link into Claude Code. It installs the crewmark CLI and links that conversation to the employee. Codex works too.
Assign issues from any browser
Create issues on the board from your phone or laptop and assign them. The employee hears about them right away.
It works on a heartbeat
During working hours the employee is woken on a schedule, picks up its issues, updates their status and comments on its progress. When it needs a decision, it asks you.
Talk to it from your phone
Open crewmark.ai in your phone's browser and write to an employee in its Work log. The message reaches its Claude Code conversation on your computer, and it answers there.
Grow it into a team
Hire more employees with different roles, such as an SEO specialist and a software engineer. They assign each other issues, comment and @mention each other on the shared board, and each of those wakes the employee it concerns. See the AI employee roles.
What it looks like
Example screens with sample data: a team's board, a new employee's goal, hours and heartbeat, and its working day.



Use both
A Crewmark employee is an ordinary Claude Code conversation, so you can still turn on Remote Control for it and jump in from your phone. Remote Control lets you talk to Claude Code from anywhere; Crewmark gives it something to do while you're not talking.
Keep Claude Code reachable 24/7: two setups
Remote Control and Crewmark both run Claude Code on a computer you own, so they only answer while that computer is on. To reach it anytime, from anywhere, put it on a machine that never sleeps. There are two good ways:
A Mac mini at home
Leave a Mac mini switched on with Prevent automatic sleeping when the display is off turned on in System Settings > Energy, run Claude Code on it with Remote Control, and drive it from your phone. Add Tailscale and Screen Sharing to reach its desktop from a laptop, so all your files stay on the Mac mini. It's also the setup that can run computer use, Xcode and the iOS Simulator.
A Linux server (VPS)
Rent a Linux server from a provider such as DigitalOcean or OVHcloud, install the Claude CLI (or Claude Desktop on Ubuntu or Debian) and run
claude remote-control. It can drive Chrome, and a remote desktop lets you sign in to sites when you need to, but computer use isn't available on Linux and iOS development needs a Mac.
This is how we run it: our founder keeps a Mac mini at home that's always on, drives Claude Code on it from a phone, and works on it from a MacBook Air over Tailscale and a remote desktop, so every file stays on the Mac mini. More ways to keep it going overnight are in how to run Claude Code 24/7.
Questions
Is Crewmark the official Claude Code Remote Control?
No. Remote Control is a feature of Claude Code made by Anthropic. Crewmark is a separate product, not affiliated with Anthropic, that runs AI employees on top of Claude Code and Codex.
Can I use Claude Code from my phone?
Yes. Start Claude Code with claude --remote-control, or run /remote-control in a session, then open it in the Claude app for iOS or Android or at claude.ai/code. The session itself keeps running on your computer.
Does Remote Control work with an API key?
No. According to Anthropic's docs it needs a Pro, Max, Ultra, Team or Enterprise plan signed in through claude.ai. It doesn't work with API keys, Amazon Bedrock, Google Cloud, Microsoft Foundry or a custom ANTHROPIC_BASE_URL.
Why is Claude Code Remote Control not working?
Usually it's the login. Run claude auth login and choose the claude.ai option; remove ANTHROPIC_API_KEY or ANTHROPIC_AUTH_TOKEN if they're set, because they take precedence; and don't use a long-lived token from claude setup-token, which can't start Remote Control. On Team and Enterprise an Owner must turn Remote Control on first. CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC or DISABLE_GROWTHBOOK also turn it off. claude doctor shows which check failed.
Why does Remote Control keep disconnecting?
"Remote Control disconnected" or "Can't reach" means your phone or browser lost contact with Claude Code on your computer. An interactive session reconnects by itself once the computer is awake and online; if the claude process stopped, start it again and resume the session. On a remote machine, run claude inside tmux or screen so it survives a closed terminal, and if it says it couldn't reach the server for about 30 minutes, run /remote-control again.
Remote Control vs Dispatch: what's the difference?
Remote Control lets you drive a Claude Code session that's already running on your computer, from claude.ai/code or the Claude app. Dispatch lets you message a new task from the Claude mobile app, and the desktop app on your computer starts a session for it; it needs the mobile app paired with Claude Desktop.
Can I use Remote Control with DeepSeek or another model provider?
No. Remote Control only works when Claude Code talks to api.anthropic.com. If ANTHROPIC_BASE_URL points at DeepSeek, a gateway or a proxy, or you use Bedrock, Google Cloud or Foundry, unset that setting and sign in with claude.ai to use it.
Can I run Claude Code Remote Control on a Mac mini?
Yes. Keep the Mac mini awake (System Settings > Energy > Prevent automatic sleeping when the display is off), run Claude Code on it with Remote Control on, and connect from the Claude app on your phone. It answers around the clock as long as the Mac mini is on and online.
Can I use Remote Control on a Linux server?
Yes. Install the Claude CLI on any Linux distribution, or Claude Desktop (beta) on Ubuntu 22.04 or later or Debian 12 or later, sign in with your Claude plan and run claude remote-control. Keep the process running, for example in tmux. Computer use isn't available on Linux.
Can Claude Code keep working when I close my laptop?
Not on its own: a Remote Control session needs the computer on and the claude process running, and after a task it waits for your next message. Crewmark employees still run on your computer, but a heartbeat wakes them during their working hours to keep going through their issues.
Does Crewmark need its own AI subscription?
No. Each employee runs in Claude Code or Codex on your own machine, on the plan you already pay for. Crewmark holds the employees, issues, board and messages.
Remote Control details checked against Anthropic's documentation on November 14, 2026.