Claude Code Remote Control not working? Fixes

When Claude Code Remote Control is not working, the message it prints names the cause. Most of the messages point at one of three things: something other than your claude.ai login is in charge (an API key, a gateway in ANTHROPIC_BASE_URL, Bedrock), a Team or Enterprise Owner hasn't turned it on, or the claude process that owns the session has stopped.

Before anything else, run claude update, then claude doctor. Anthropic's Remote Control docs say claude doctor shows which eligibility check failed, and several messages changed wording between versions, so an old build can print a message that no longer matches the docs.

Remote Control error messages and their fixes#

The wording below is from Anthropic's Remote Control troubleshooting section and error reference, read on 24 October 2026.

Why it says "requires a claude.ai subscription" on a paid plan#

When an API key or token is set, Claude Code uses it ahead of your claude.ai login. Anthropic's docs say the signed-in version of the message names the credential, for example ANTHROPIC_API_KEY is set, so this session is using API-key auth. Remote Control never accepts API-key auth, so the fix is to take that key out of the session's way, not to sign in again.

1. Run env | grep ANTHROPIC in the terminal where you start Claude Code. If it lists ANTHROPIC_API_KEY or ANTHROPIC_AUTH_TOKEN, unset it and find the line in your shell profile that sets it.
2. Open ~/.claude/settings.json, plus .claude/settings.json and .claude/settings.local.json in the project, and remove the variable from each env block. Remove an apiKeyHelper line too.
3. Start a new session and run /remote-control again.

The same goes for "Remote Control is only available when using Claude via api.anthropic.com". The message names what routed you away, such as CLAUDE_CODE_USE_BEDROCK or a custom ANTHROPIC_BASE_URL, and the error reference adds that a gateway blocks Remote Control "even when you sign in with claude.ai". If your company requires the gateway, Remote Control is off the table for those sessions. Claude Code on Bedrock or Vertex covers that setup.

Remote Control isn't enabled for this account, or your organisation blocks it#

"Remote Control isn't enabled for this account" is about the account itself: the check came back off. Anthropic's docs put it down to cached entitlements after a plan change, fixed by claude auth logout, then claude auth login, and an update if your Claude Code is old. Before v2.1.239 the same message read "Remote Control is not yet enabled for your account".

"Disabled by your organization's policy" is a different check. On Team and Enterprise plans, Remote Control is off by default, and an Owner turns it on with the Remote Control toggle at claude.ai/admin-settings/claude-code. If it works for colleagues but not for you, and your Enterprise org uses custom roles, you need both the cloud sessions permission and the Claude Code remote control permission. If the message mentions disableRemoteControl, your IT administrator turned it off on this device through managed settings, whatever the org toggle says. An organisation with Anthropic's HIPAA configuration can't turn it on at all; /status shows HIPAA on its Organization configuration line.

There is a trap for people who left a company plan. If you're on Pro or Max now but Claude Code still holds a login from an earlier Team or Enterprise organisation, it checks that organisation's policy. Run /status to see which plan and org you're signed in under, then claude auth logout and claude auth login.

If you set DISABLE_TELEMETRY or DO_NOT_TRACK, Remote Control still works on Claude Code v2.1.283 or later (v2.1.304 in the Desktop app), unless your organisation requires Trusted Devices. Then you have to unset them. CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC and DISABLE_GROWTHBOOK turn it off on any version.

Remote Control keeps disconnecting#

When the network drops or changes, what Claude Code does next depends on how you started Remote Control.

After a VPN or network change, something between your machine and Anthropic can start answering with HTTP 403. An interactive session keeps retrying for up to three minutes, then disconnects and names what refused: a network edge, or a proxy, VPN or firewall on your side.

When the machine is awake but has no network, the two modes part ways. claude remote-control (server mode) gives up after roughly 10 minutes and exits, so you run it again. An interactive session retries for as long as the outage lasts and reconnects by itself.

When the laptop sleeps, Claude Code reconnects once it wakes and is online, but nothing runs in between. Our lid-closed guide has the free ways to keep a MacBook awake and what each one risks.

Started Claude Code before the VPN connected? That gives "Couldn't verify your organization's policy". Each new /remote-control checks again, so you don't need to restart Claude Code.

If an Owner switches Remote Control off while you're connected, the session drops with "turned off by your organization's policy" and doesn't come back on its own. A session started with /remote-control or claude --remote-control keeps running locally; claude remote-control (server mode) archives its sessions and exits.

The session went offline: the claude process stopped#

Remote Control is a window into a process on your computer. Close the terminal, quit the Desktop app or VS Code, or stop claude any other way, and the session goes offline. On a remote machine over SSH, the docs say to start claude inside tmux or screen so it survives the SSH connection dropping. The Claude Code over SSH guide walks through that setup, and running several sessions in tmux covers the layout.

How to bring a stopped session back#

1. If you stopped claude remote-control with Ctrl+C, go to the same directory and run claude remote-control to bring back every session it served, claude remote-control --continue for only the session it started with, or claude remote-control --session-id <id> for one. The ID is the part of the claude.ai/code URL between /code/ and any ?. These work for about four hours after the server stopped, unless you started it with --no-create-session-in-dir or ran a second claude remote-control in the same directory; --continue and --session-id need Claude Code v2.1.200 or later.
2. If you started it with claude --remote-control or /remote-control, resume the conversation with claude --continue or claude --resume. If that prints "Couldn't reconnect to your Remote Control session", run /remote-control.
3. If Claude Code says "Previous session is unavailable", run /remote-control to start a new remote session under your current login.

Remote Control drives a session on your own machine. A session that runs in Anthropic's cloud is a different feature, with its own errors: pulling one into your terminal is covered in Claude Code teleport. Remote Control works only for Claude Code; Codex has its own, through the ChatGPT app, and running Codex from your phone has its setup and its failure modes.

How Shuriken does it#

Shuriken doesn't go through Remote Control, so none of the messages above apply to it. Your agents run on your own computers and servers, and with the background service installed (Settings › Devices) they keep running after you close the app window. The web app at app.shuriken.sh follows every chat on every computer and server you've added, whatever the agent: Claude Code, Codex, Gemini CLI or any of the others Shuriken runs. What travels between your devices is end-to-end encrypted, and our cloud relays it without being able to read it.

As with Remote Control, the computer has to be on. A computer signed in to Shuriken starts with Keep awake set to Always, and Cocaine mode keeps a Mac running with the lid closed. A Shuriken server runs as a service and restarts on its own. Native iPhone and Android apps are planned for November 2026; until then, the phone app is the web app, and our guide to running Claude Code from your phone compares all four routes. To try it on your own machines, download Shuriken.