Claude Code Remote Control not working? Fix it by symptom
You typed the command, and your phone shows nothing, or the session was there a minute ago and now it is gone. Almost every case comes down to four causes.
Short answer: Claude Code Remote Control fails most often because you are not signed in with a claude.ai plan login, an API key or custom endpoint is overriding it, or the computer running the session went to sleep or lost its network. Check your login first with claude auth login, then keep the machine awake, and look for the session under the Code tab of the Claude app on the same account.
Remote Control is Anthropic's feature for steering a Claude Code session that runs on your own computer from the Claude app on your phone or from claude.ai/code. Everything below about how it works comes from Anthropic's Remote Control documentation, read on October 19, 2026. Where I cite a user report, it is a public issue on the claude-code GitHub repository, and I say what state it was in when I read it.
First, what Remote Control actually is
Per the docs, Claude keeps running on your machine the whole time. Your phone or browser is a window into that local session, so "your computer has to stay on and the claude process has to keep running." That one sentence explains most of the "disconnects" complaints further down.
The session makes outbound HTTPS requests only and never opens inbound ports on your computer. It registers with the Anthropic API and polls for work. That is why it can work through a home router with no setup, and also why a firewall, proxy or VPN that blocks that outbound traffic breaks it.
The command, and the three ways to start it
The docs list three CLI modes. Pick one and make sure you are in the project folder you want to work on.
- Server mode: run
claude remote-control. The process stays running in your terminal and shows a session URL. Press the spacebar to show a QR code you can scan with your phone. - Interactive session: run
claude --remote-control(or--rc) to get a normal terminal session that is also reachable remotely. - From a session you already started: type
/remote-control(or/rc). It carries your current conversation over.
The first time you run it, Claude Code asks you to confirm. Decline and it asks again next time. In a folder you have not trusted yet, claude remote-control also asks Trust <directory>? [y/N]. If the command is running somewhere that cannot show a prompt, it exits with a "Workspace not trusted" error instead.
Symptom 1: "not available", "not enabled" or "requires a claude.ai subscription"
This is the biggest group. The docs list these requirements, and each error message maps to one of them.
- Plan: Pro, Max, Team and Enterprise. API keys are not supported. On Team and Enterprise, an Owner has to switch on the Remote Control toggle in Claude Code admin settings first, because it is off by default there.
- Login: sign in through claude.ai. Run
claude auth loginand choose the claude.ai option. If the message saysANTHROPIC_API_KEY,ANTHROPIC_AUTH_TOKENorapiKeyHelperis in use, remove it from your shell and from theenvblock of your settings file. - Token type: a long-lived token from
claude setup-tokenorCLAUDE_CODE_OAUTH_TOKENcannot start Remote Control. The docs say it can only make model requests. Runclaude auth loginfor a full session token. - Endpoint: it does not work on Amazon Bedrock, Google Cloud's Agent Platform or Microsoft Foundry. It also fails if
ANTHROPIC_BASE_URLpoints at anything other than api.anthropic.com, including an LLM gateway or proxy, even with a valid claude.ai login. Unset the variable. - Privacy variables:
CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFICorDISABLE_GROWTHBOOKturn Remote Control off.DISABLE_TELEMETRYorDO_NOT_TRACKalone are fine from v2.1.283 onward, and in versions v2.1.154 through v2.1.282 they produced the error.
The error "Remote Control isn't enabled for this account" has its own fix in the docs: the usual cause is stale cached entitlements after a plan change. Run claude auth logout, then claude auth login, and update Claude Code. Run claude doctor to see which individual eligibility check failed. If you are on Pro or Max but still get a policy error, the docs say you may still be signed in under an old Team or Enterprise organization. Run /status to see which one, then log out and back in.
About the version: the Requirements section of the docs does not list a minimum Claude Code version. The version only matters through specific messages. The wording of that "isn't enabled" error changed in v2.1.239, and "unexpected server response" is fixed by running claude update. So when in doubt, update first.
You are not the first to hit this. Issue #35322 is a user on Claude Code 2.1.58 who got "Remote Control is not enabled for your account" while titling it a Pro plan problem. When I read it, it was closed as a duplicate and locked, with no maintainer reply on the page. I would treat it as proof that the symptom is common, not as a diagnosis.
Symptom 2: it disconnects when the laptop sleeps
The docs say that if your laptop sleeps or the network drops, Claude Code reconnects automatically when the machine comes back online. They also say that if you close the terminal or stop the claude process, the session goes offline until you bring it back. A sleeping laptop is not a running process doing work, so plan around that.
The straightforward fix is to stop the Mac from sleeping while the session matters:
- Open System Settings. On a MacBook go to Battery, then Options, and turn on "Prevent automatic sleeping on power adapter when the display is off" (Apple's sleep and wake settings page). On a desktop Mac it is under Energy, with the same idea.
- Or run
caffeinate -iin a second terminal tab. Apple'scaffeinateman page says it creates an assertion to prevent the system from idle sleeping, and the assertion lasts until caffeinate exits. The-sflag is stronger but, per the man page, works only on AC power. - Leave the lid open and the charger in. These steps address idle sleep. I am not claiming they keep a laptop running with the lid closed, because I did not find that in the sources above.
- If you run Claude Code on another machine over SSH, start it inside tmux or screen, as the docs recommend, so closing your SSH window does not end the session.
Be honest with yourself about one thing: sleep settings do not fix every drop. Issue #33041 (open when I read it) is a macOS user on 2.1.72 whose session dropped every few minutes even with sleep disabled through caffeinate and a VPN connected. Many others added similar reports through September 2026 on macOS, Windows and Linux. There was no maintainer reply on the page. And #34531, closed as a duplicate, describes a session that lost sync after a disconnect and could not be restored with /rc; the only workaround the reporter found was a new session.
What that means in practice: if /remote-control does not bring a dropped session back, start a fresh one. The docs also describe --continue and --session-id in server mode to bring a stopped session back, and both need Claude Code v2.1.200 or later.
Symptom 3: network, proxy and VPN problems
The docs are specific here, which helps.
- Port 443 to the Anthropic API has to be open. A firewall or proxy that blocks it produces "Remote credentials fetch failed". Rerun with
claude remote-control --verboseto see the real error. - VPN or network switch mid-session: if something answers with HTTP 403, an interactive session keeps retrying for up to three minutes. After that it disconnects and names what refused, either a network edge or a proxy, VPN or firewall on your side.
- Starting before the VPN is up: "Couldn't verify your organization's policy for remote control" usually means you started offline or before a VPN connected. Restore the network, then run
/remote-controlagain. - A long outage: in server mode, Claude Code gives up after roughly 10 minutes and the process exits, so you run
claude remote-controlagain. An interactive session keeps retrying for as long as the outage lasts. - A 30 minute failure: if you see "could not reach the Remote Control server for about 30 minutes", run
/remote-controlto reconnect.
Symptom 4: the session does not show up on your phone
Work through this in order. Each point comes from the docs.
- Check the terminal first. In an interactive session you should see an
/rc activeindicator. If the connection failed, the indicator changes and a reason is added to the conversation. Run/remote-controlto open the status panel with the session URL and QR code. - In the Claude mobile app, tap Code in the navigation to reach the session list. Online Remote Control sessions show a computer icon with a green status dot. No Claude app yet? Run
/mobileinside Claude Code for a QR code to the right app store. - Scan the QR code or open the session URL, instead of hunting through a long list. The session title is the name you passed, or your
/renametitle, or a generated name that starts with your computer's hostname, like myhost-graceful-unicorn. - Make sure you are signed in to the same claude.ai account on the phone. The docs say a session started by auto-connect appears only in your own account's Claude apps.
- Read the failure reason. "Another connection took over this session" means another device or Claude Code session has it now. "This session was ended or archived from another device or app" means you only need to run
/remote-controlif you want it back. "The server no longer reports this session" may mean it was deleted elsewhere. - Look for archived sessions. The docs link a filter for archived sessions on claude.ai/code, which is where a session ends up when an organization policy switches Remote Control off mid-session.
If the session shows up but your messages do nothing, you are in a different bug class. Issue #36008 is a macOS user on 2.1.58 whose session appeared in the app but whose messages never reached the CLI. The reporter closed it themselves, and no root cause was posted. The docs do not describe this case, so the practical moves are to update with claude update and start a fresh session.
What this costs you, and the limits
I found no separate price for Remote Control in the docs. It is a Claude Code feature available on the plans listed above, so if you already pay for one of them, it is included. Treat that as "no extra charge I could find", not a promise. Check your own plan page.
- Your computer must stay on, awake, online, with the claude process running. There is no way around that in the design.
- Outside server mode, each Claude Code instance supports one remote session at a time.
- While connected, the session transcript, including your messages, Claude's replies and tool activity, is stored on Anthropic's servers so it can sync across devices and reconnect after a drop. Execution and file access stay on your machine.
- Organizations with Zero Data Retention or the HIPAA configuration cannot enable it.
- Some commands only work in the terminal, such as
/pluginand/resume. - Permission prompts stay open until you answer them. Other forwarded dialogs close after five minutes by default and continue with their no-action default.
- It controls Claude Code. It is not a view of your whole screen.
If you want a different setup
Remote Control is first-party, ships with the plans above and is the most direct way to steer a Claude Code conversation from your phone. If it works for you, stay with it. Two situations make people look elsewhere, and this is my honest read of them. I build MightyMouse, so weigh this section accordingly.
The first is that you want more than the conversation. MightyMouse shows your phone the whole screen of your Mac and gives you a trackpad and keyboard, so you can see the dialog the terminal is stuck behind or open the browser a build opened. It has an Away Mode that keeps the Mac awake while you work it remotely. It cannot reach a Mac that is asleep, and it cannot get past a lock screen that is already engaged. It also only works on macOS. Control is free and unlimited on your own Wi-Fi, and away from home it runs over cellular with a free monthly allowance.
The second is that you use more than one agent. MightyMouse runs an MCP server inside the Mac app, and per its own agent access notes any agent that speaks MCP, such as Claude Code, Codex, Cursor and VS Code, can send your phone a notification, a question or an approval request. You answer from the lock screen. An unanswered approval counts as a no, and an unanswered question comes back to the agent as timed out. The push shows a short preview of the question, up to 300 characters. The limit to know: these are discrete cards, not a live line-by-line view of the terminal, so it replaces neither Remote Control's conversation view nor a screen view. The app has to be running for any of it to work.
Many people run both: Remote Control to chat with a Claude Code session, and a tool that shows the Mac for everything else. If you are comparing the field, see Claude Code on iPhone, Claude Code notifications on your phone, Happy Coder, how this compares with Claude Dispatch and the roundup of every way to run coding agents from your phone. And if the real problem is that your Mac keeps sleeping on you, read how to wake a Mac remotely.
Your phone gets the whole screen of your computer, plus a trackpad.
No card to try it.
Common questions
Why is Claude Code Remote Control not available?
Per Anthropic's docs, it needs a Pro, Max, Team or Enterprise plan and a claude.ai login, not an API key. It also fails with Amazon Bedrock, Google Cloud's Agent Platform, Microsoft Foundry, or a custom ANTHROPIC_BASE_URL. On Team and Enterprise an Owner must switch it on first. Try claude auth logout, then claude auth login, and run claude doctor.
What is the Claude Code Remote Control command?
Run claude remote-control for server mode, claude --remote-control for an interactive session, or type /remote-control inside a session you already started. In server mode, press the spacebar to show a QR code for your phone.
Why does Claude Code Remote Control keep disconnecting?
The session runs on your computer, so sleep, a closed terminal, a VPN change or a network outage can drop it. The docs say it reconnects when the machine is back online, and server mode gives up after about 10 minutes offline. Some users report drops even with sleep disabled, in public GitHub issues.
Does Claude Code Remote Control work when my laptop is asleep?
No. The local claude process has to keep running. Anthropic says it reconnects after sleep once the machine is back online, but nothing runs while the laptop sleeps. Change your sleep settings or use caffeinate to prevent idle sleep.
How do I use Claude Code Remote Control from my phone?
Start a session with claude remote-control or /remote-control, scan the QR code or open the session URL, or open the Claude app and tap Code. Sessions that are online show a computer icon with a green dot. Sign in with the same claude.ai account on both devices.
Is Claude Code Remote Control free?
It is a Claude Code feature on Pro, Max, Team and Enterprise plans. The docs list no separate charge for it, but they do not say it is free either, so check your own plan.
If you want your phone to see and drive the whole computer, not only one session: MightyMouse puts your Mac on your iPhone as a live screen, trackpad and keyboard, and relays questions from any MCP agent to your lock screen.