Short version. Remote Control needs a claude.ai login on a Pro, Max, Team or Enterprise plan, a current Claude Code, and a claude process still running on a computer that is awake. claude doctor shows which eligibility check fails. If it connected and then dropped, read the reason Claude Code printed, then run /remote-control again.

How Remote Control works

Per Anthropic's docs, Remote Control connects claude.ai/code or the Claude app for iOS and Android to a Claude Code session running on your machine. Claude keeps running locally, with your files, tools and MCP servers. The phone is a window into that session, so your computer has to stay on and the claude process has to keep running.

The local session makes outbound HTTPS requests only and opens no inbound ports. Anthropic's servers route messages between your phone and the session. While it is connected, the transcript is stored on Anthropic's servers; execution and file access stay on your machine.

You start it one of three ways:

  • claude --remote-control (or --rc): a normal interactive session that is also reachable remotely.
  • /remote-control (or /rc) inside a session you already have open.
  • claude remote-control: server mode, which waits for remote connections and can serve several sessions.

Then open the session URL it prints, scan the QR code, or find the session in the Claude app under Code.

It won't start, or says it isn't available

Most likely first. The quoted text is what Claude Code prints, per the docs.

  1. You aren't signed in with a claude.ai account, or an API key is taking over. The message is some form of "Remote Control requires a claude.ai subscription", or "requires claude.ai subscription auth" followed by the credential in use, such as ANTHROPIC_API_KEY. API keys are not supported. Run claude auth login, choose the claude.ai option, and remove ANTHROPIC_API_KEY or ANTHROPIC_AUTH_TOKEN from your shell environment and the env block of your settings file.
  2. Your organisation hasn't turned it on. "Remote Control is disabled by your organization's policy". On Team and Enterprise plans it is off by default until an Owner enables the Remote Control toggle in the Claude Code admin settings. On Pro or Max, this message means Claude Code is still signed in under an old organisation: check with /status, then log out and in again.
  3. Stale account information. "Remote Control isn't enabled for this account" usually means cached entitlements are out of date after a plan change. Run claude auth logout, then claude auth login.
  4. You aren't talking to Anthropic's API directly. Remote Control is not available on Amazon Bedrock, Google Cloud or Microsoft Foundry, or when ANTHROPIC_BASE_URL points at a gateway or proxy. The error names the variable to unset.
  5. A privacy variable is set. CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC or DISABLE_GROWTHBOOK makes Remote Control unavailable. DISABLE_TELEMETRY and DO_NOT_TRACK did the same before v2.1.283.
  6. An old version. The docs tie many fixes to specific 2.1.x releases, and "Remote Control got an unexpected server response" is fixed only by updating. Run claude update.
  7. No network at startup. "Couldn't verify Remote Control eligibility" and "Remote credentials fetch failed" usually mean Claude Code couldn't reach Anthropic: you are offline, or a proxy or firewall is blocking outbound HTTPS on port 443. Retry once you are online.

It connects, then disconnects

When the connection fails in an interactive session, Claude Code shows the reason in a notification and adds it to the conversation, on a line that starts with Remote Control disconnected. Read that first. The documented causes:

  • The process stopped. Closing the terminal, quitting the Desktop app or VS Code, or stopping claude takes the session offline. If you stopped server mode with Ctrl+C, running claude remote-control again in the same directory brings its sessions back for about four hours. On a machine you reach over SSH, the docs say to start Claude Code inside tmux or screen so it outlives the SSH connection.
  • The computer went to sleep. Claude Code reconnects on its own when the machine comes back online, but nothing answers while it is asleep, and your phone can't wake it. On a Mac, turn on Prevent automatic sleeping on power adapter when the display is off and keep it plugged in; our Tailscale SSH guide says where that setting lives.
  • The network changed. After a VPN or network change, something between you and Anthropic can start answering with HTTP 403. Claude Code retries for up to three minutes, then disconnects and names what refused. In a longer outage, an interactive session keeps retrying for as long as it lasts; server mode gives up after roughly ten minutes and exits.
  • Your login expired. The connection runs on short-lived credentials renewed from your saved login. If the reason mentions Claude.ai login expired or an OAuth token, run /login, then /remote-control.
  • Another device took the session. "Another connection took over this session" means what it says. Run /remote-control only if you want it back.

None of those, and it still drops

On Anthropic's issue tracker, issue #33041 has been open since March 2026 and was still collecting reports in late September 2026 from people whose machines never sleep and whose networks are stable. #31853 describes a drop roughly every 25 minutes on a wired Mac mini, and #50767 a disconnect after about two hours; both were closed as not planned. These are user reports, not confirmed causes, so update first. The workaround people report is the documented one: type /remote-control again in the session. The catch is that you have to reach the computer to type it.

When Remote Control is the right choice

If you have a Claude subscription and only use Claude Code, start with Remote Control. There is nothing to install beyond the Claude app, no SSH, no port to open, and it is the supported path: messages and permission prompts sync both ways, and the app can send push notifications.

It fits less well when you use an API key or a cloud provider, when your organisation has disabled it, when you also want your shells, dev servers or other agents, or when you need to recover a dropped session without walking to the desk. Some commands, such as /plugin and /resume, only work in the local terminal, and outside server mode each Claude Code process supports one remote session.

The alternative: a multiplexer and SSH

The other route is to run Claude Code inside a terminal multiplexer on your computer and reach it from the phone over SSH, usually through Tailscale. It is more setup and you maintain it yourself. In exchange it doesn't depend on your plan or login type, and a dropped connection costs nothing: the session keeps running and you attach again from the phone. It also covers the gap above, because once you can SSH in, you can type /remote-control into the session from anywhere.

How to run Claude Code from your phone walks through the whole setup. Melta, our app for this route, is coming soon to the App Store and Google Play: it connects over SSH to Herdr, a multiplexer built for coding agents, and reconnects on its own after a drop. It can't wake a sleeping computer either.

Common questions

How do I enable Claude Code Remote Control? Run claude --remote-control, or type /remote-control in an open session, and accept the one-time confirmation. To connect every session automatically, set Enable Remote Control for all sessions in /config.

Why is Remote Control not available on my account? It needs a claude.ai login on a Pro, Max, Team or Enterprise plan. It isn't available with API keys, on Bedrock, Google Cloud or Foundry, through a custom ANTHROPIC_BASE_URL, or for organisations with Zero Data Retention requirements.

Does Remote Control work from a phone? Yes. Use the Claude app for iOS or Android and tap Code, or open claude.ai/code in the phone's browser.

Does Remote Control keep working if I close my laptop? No. The session runs on that laptop. It reconnects when the machine wakes, but nothing happens while it sleeps.

Sources