What does "/rc active" mean in Claude Code?
The /rc active badge in your Claude Code footer means Remote Control is connected, so the session running on your machine can be driven from claude.ai or the Claude app. What turned it on, what it does with your data, what to do when it turns red, and how to switch it off.
Short answer: /rc active in the footer below your prompt box means Remote Control is connected. Your Claude Code session is still running on your own machine, and it can now also be driven from claude.ai/code or the Claude mobile app. Nothing has moved to the cloud; the phone is a window into the session on your laptop.
That is the whole thing. The rest of this post is the questions people have right after: why it appeared without them asking, whether it matters for privacy, what to do when it turns into an error, and how to turn it off.
Why it showed up
Remote Control activates in exactly four ways:
- You ran
claude remote-control(server mode, no local prompt). - You started the session with
claude --remote-controlorclaude --rc. - You typed
/remote-controlor/rcmid-session. - Auto-connect is on, which connects every interactive session automatically.
If you did not do any of the first three, it is the fourth. The toggle lives in /config as Enable Remote Control for all sessions, and it is worth knowing that a managed settings value of true from your organization outranks a false you set for yourself. The exception: a false in project or local settings (.claude/settings.json, .claude/settings.local.json) turns auto-connect off even over a managed true.
The indicator hides itself when the terminal is too narrow, so a missing badge does not always mean a disconnected session.
What you can do with it
The badge is a link to the session on claude.ai. Select it with the down arrow key and press Enter, or just run /remote-control again, and you get a status panel with:
- the session URL, to open on another computer,
- a QR code, to open it in the Claude app on your phone,
- a disconnect option.
The practical use is walking away mid-task. A long turn finishes while you are on the couch and you read the result on your phone; a permission prompt appears and you approve it without going back to the desk. Claude Code nudges you toward this at the two moments it helps most, showing a Still working, check in from your phone notice during a long turn, and an Approve tool calls from your phone notice after you have answered several permission prompts. Those are not configurable and clear on their own.
Should you worry about it?
The honest version, because "something connected without me asking" is a fair thing to want details on:
- Your code and commands stay on your machine. Remote Control runs the session locally; the web and mobile interfaces are remote controls, not remote computers. This is the opposite of Claude Code on the web, which does run in Anthropic's cloud.
- No inbound ports open. Your machine makes outbound HTTPS requests only, registering with the Anthropic API and polling for work. Traffic runs over TLS with multiple short-lived credentials, each scoped to one purpose.
- The transcript is stored on Anthropic servers while connected, because that is how the conversation stays in sync across your devices. That is the part worth knowing if your threat model cares about where conversation text lives.
- It requires a Pro, Max, Team, or Enterprise plan, and it does not work with API keys, Bedrock, Google Cloud, Microsoft Foundry, or a custom
ANTHROPIC_BASE_URL.
Organizations can also require Trusted Devices, a Face ID, Touch ID, Windows Hello, or hardware key check before a device is allowed to connect. The wider picture is in our Claude Code security writeup.
When the badge shows a failure
If the connection drops, the indicator switches to a failure state and stays there. Select it with the down arrow and press Enter to read the reason, and read it before you reconnect, because three of the reasons mean something other than "the network hiccuped":
| Reason | What it means | What to do |
|---|---|---|
| Connection or heartbeat failure | The usual network case | /remote-control to reconnect |
| Another device or session took it over | Something else is driving this session now | Reconnect only if you want it back from that device |
| You ended or archived it elsewhere | You closed it from the app or another session | Reconnect only if you want it back; an archived session reopens |
| The server cannot find the session | It was probably deleted from another device | Start a new one |
Two timeouts explain most disappearances: server mode gives up after roughly ten minutes of failed reconnection and exits, and an interactive session reports a disconnect after about thirty minutes of failing presence heartbeats.
One more thing that surprises people: the local process has to stay alive. Sleeping your laptop is fine, since it reconnects on wake, but closing the terminal or quitting VS Code takes the session offline. If your agent runs on a remote box over SSH, run it inside tmux, or a dropped connection kills the process and Remote Control with it.
How to turn it off
- This session: open the status panel (down arrow on the badge, then Enter, or
/remote-control) and choose disconnect. Your local session keeps running in the terminal. - Every session:
/config, then set Enable Remote Control for all sessions to false. Remember the precedence rule above if your organization pushed it on.
Related
Remote Control answers "I am away from the desk". It does nothing for the more common case of being at your desk in a different window, which is what finish and blocked notifications are for, and it does not make the waiting itself shorter, which is the problem Unwait exists for. The full feature writeup, including the three ways to start it and how it compares to Claude Code on the web, is in Claude Code Remote Control: what it replaces, and what it does not.