◉CLAUDE LABJP
●CC 2.1.325 — No new release over the weekend; 2.1.325 (11/13) is still the latest●BILLING — 2.1.325 fixes background sessions on a claude.ai login being billed to an API key from the shell●SONNET 4.5 — Retires on the Claude API on 11/30, 14 days left. Check saved model settings too●403 — Users report a Cloudflare 403 page mid-session that still burns tokens●MANAGED — Since 11/12 you can request self-hosted Claude Managed Agents that run entirely on your own infrastructure●NEW — When Browser Extension Is Not Connected Won't Go Away●CC 2.1.325 — No new release over the weekend; 2.1.325 (11/13) is still the latest●BILLING — 2.1.325 fixes background sessions on a claude.ai login being billed to an API key from the shell●SONNET 4.5 — Retires on the Claude API on 11/30, 14 days left. Check saved model settings too●403 — Users report a Cloudflare 403 page mid-session that still burns tokens●MANAGED — Since 11/12 you can request self-hosted Claude Managed Agents that run entirely on your own infrastructure●NEW — When Browser Extension Is Not Connected Won't Go Away
Articles/Claude Code
⟐ Claude Code/2026-08-15Beginner

If You Walk Away While Claude Code Is Running, Turn On Remote Control First

Hand off a long task, step out, and you often come back to a session that stalled at the very first approval prompt. Remote Control lets you reach the Claude Code session running on your own machine from your phone or a browser. Here is how to set it up and what to know before you rely on it.

Claude Code285Remote ControlSession Management4Mobile

I once handed off a batch of dependency updates and left to run errands.

Thirty minutes later I came back to find Claude Code waiting quietly at the very first approval prompt. Actual work done in that half hour: maybe a minute's worth.

If you are an indie developer shipping apps on your own, this kind of "time spent only waiting" adds up fast. The decision itself takes three seconds, but those three seconds require you to be sitting at your desk. After it happens a few times, you start avoiding long-running handoffs altogether.

Remote Control exists to close exactly that gap. It landed alongside claude self-hosted-runner in v2.1.224 on August 7, 2026, and v2.1.229 on August 12 added --continue for picking a stopped session back up. It is available on Pro, Max, Team, and Enterprise subscriptions, and not with API-key authentication (I checked the official docs again on October 23, 2026).

Remote Control does not move your work to the cloud

Let me clear up the most common misreading first.

Turning Remote Control on does not relocate anything to a cloud environment. Claude Code keeps running on your machine. Your phone and your browser simply become windows into that local session.

Because of that design, a few things stay true:

  • File reads and writes happen against your own filesystem
  • Your configured MCP servers and project-specific settings remain available
  • Typing @ on your phone autocompletes file paths from your local project

So there is no "it behaves differently because it re-ran somewhere else" problem. It is closer to peeking at the Mac on your desk from across town and adding an instruction. The conversation stays in sync across terminal, browser, and phone, so you can switch between them mid-task without losing the thread.

The flip side is equally direct: if your machine is off, nothing happens. That one is easy to trip over if you assume the work lives in the cloud.

The shortest path to your first session

You need three things:

  1. Claude Code v2.1.224 or later
  2. A claude.ai subscription
  3. Authentication with a claude.ai account

If you currently authenticate with an API key, you will need to sign in again with the claude.ai option. A leftover ANTHROPIC_API_KEY in your environment can make Claude Code treat the session as API-key auth even after you sign in, so check that too. Setups that reach Claude through Amazon Bedrock and similar clouds are out, because Remote Control needs a direct connection to the Anthropic API.

# Check your version
claude --version
 
# Sign in with a claude.ai account (pick the claude.ai option)
claude auth login

Once that is done, start a session with Remote Control attached.

# Starting a fresh session
claude remote-control
 
# Attaching to a session that is already running,
# from inside Claude Code
/remote-control

Then open the Claude app on your phone and tap Code in the navigation, or open the Code view on claude.ai in a browser. Your running sessions are listed there. Tap one and you pick up the conversation where it stands. A connected session shows /rc active in your terminal, and running /remote-control again prints the session URL and a QR code; scanning that with your phone camera is the quickest way in.

If typing the command every time gets tiresome, you can leave it on permanently.

# Open /config inside Claude Code and enable
# "Enable Remote Control for all sessions"
/config

One note for teams: on Team and Enterprise plans this is off by default until an Owner flips the Remote Control toggle in the Claude Code admin settings. Personal accounts have no such gate.

What to check before you walk away

These are the details that tend to surprise people mid-errand. Knowing them in advance saves a small panic on the train.

AreaWhat applies
Permission modesSet the starting mode with --permission-mode when you launch the server (for example acceptEdits; manual is an alias for default). The docs do not spell out which modes you can switch to from the phone, so I would try it once at your desk before relying on it
Local-only commands/plugin and /resume run from the terminal only, with or without Remote Control
NetworkYou need to reach the Anthropic API on port 443; a firewall or proxy that blocks it will stop the connection
InterruptionsIf your laptop sleeps or the connection drops, Claude Code reconnects automatically once the machine is back online
Long outagesIf the machine is awake but offline, claude remote-control (server mode) gives up after roughly ten minutes and the process exits. An interactive session keeps retrying and reconnects by itself when the network returns

What shapes daily use most is that approval requests reach your phone more often than you expect. If you normally let auto mode run unattended, the number of confirmations when you are away may surprise you.

So plan Remote Control work on the assumption that you are in the loop. If you shape the task so there are fewer approval points before you leave, you can handle the rest one-handed while you are out.

Outages behave differently by mode. Server mode exits after about ten minutes, so wrapping up a checkpoint before you head underground is a small habit that pays off. An interactive session simply reconnects once you are back above ground.

One more thing: what runs is a local process. Close the terminal, or quit VS Code or the desktop app hosting the session, and it goes offline. If you run claude over SSH, start it inside tmux or screen so it survives a dropped connection.

Picking a stopped session back up

When you stop claude remote-control with Ctrl+C, the sessions it was serving stop responding from your phone or browser. They are not archived, though, as long as you were not running another claude remote-control in the same directory and did not start this one with --no-create-session-in-dir.

There are three ways back, and they differ in how much they restore.

CommandWhat comes backWhen to use it
claude remote-controlEvery session that server was servingYou had several tasks running in parallel
claude remote-control --continueOnly the session the server started with; it exits when that session endsOne task, and you just want the thread back
claude remote-control --session-id <id>Only the session whose ID you passYou already know which one you want
# Bring back just the session this server started with
claude remote-control --continue
 
# Bring back one specific session by ID
claude remote-control --session-id <session-id>

There is one deadline worth memorizing: these work for about four hours after the server stopped. Past that, run claude remote-control to start fresh. claude --resume and claude --session-id follow the same four-hour window.

If you archived a session in the meantime, --continue and --session-id will unarchive it on v2.1.228 or later. That is a meaningful improvement over having to hunt for it manually.

A session you started with claude --remote-control or /remote-control also comes back with Remote Control still attached when you resume the conversation using claude --continue or claude --resume. The condition is that Remote Control was still on when the session ended.

When a session refuses to restore at all, the cause is usually session management rather than anything Remote Control specific. In that case, start with what to do when Claude Code sessions won't restore.

Always on, or only when you need it?

You can enable it for every session through /config, but I keep it manual and deliberate.

The reasoning is plain: a session with Remote Control on is a session reachable from outside. For work that finishes in a couple of minutes at my desk, there is no reason to open that window at all.

I do turn it on before stepping away when the task looks like one of these:

  • Bulk dependency updates, or anything with a long build in it
  • Work I know will hit approval points a few times
  • Work where I want to see a result before deciding what comes next

Before I leave I ask myself one thing: will Claude need my judgment even once while I am gone? If not, skip it. If yes, turn it on. That alone removes most of the wasted trips back to the desk.

If it is the connection itself that dies partway through a long reasoning step, the cause is often the path rather than your Remote Control setup. When a gateway or proxy sits in the middle, the piece on relays killing long-running streams walks through how to isolate it.

One thing to try next

If you do one thing today, make it this: start your next long-running task with claude remote-control.

Step away for a few minutes and check that the same conversation shows up in the Claude app on your phone. Once you have seen it work, the time you spend tethered to your desk for approvals becomes a choice rather than a given.

I only recently made this part of my routine, and I am still figuring out where it fits best. What I can say is that deciding to leave the desk got noticeably easier.

Share

Thank You for Reading

Claude Lab is ad-free, supported entirely by members like you. We publish practical guides daily with implementation code, benchmarks, and production-ready patterns. If you've found it useful, we'd love to have you on board.

  • ✦Copy-paste ready implementation code
  • ✦New advanced guides published daily
  • ✦$5/mo or $10 for lifetime access
View Membership →

If you found this article helpful, a small tip ($1.50) would mean a lot to us. Your support helps keep this site ad-free and covers server and hosting costs.

Related Articles

⟐ Claude Code2026-07-18
Branch with /fork, Delegate with /subtask — Two Commands That Traded Jobs
In Claude Code 2.1.212, /fork now clones your conversation into a background session, and in-session subagents moved to /subtask. Here is how I decide between them, and the budgets worth setting before you start branching freely.
⟐ Claude Code2026-11-08
Before a Nightly Job Hangs on a Prompt: Designing Around denyInsteadOfPrompting
Claude Code 2.1.319 adds permissions.denyInsteadOfPrompting. Here is how it compares with other ways to answer permission prompts unattended, and how to catch what a denial quietly drops.
⟐ Claude Code2026-11-06
When claude doctor Listed Four Warnings, I Sorted Them by Cause Before Fixing Anything
claude doctor ended with "4 warnings found", but I didn't need four fixes. Using the real output, here is how to spot duplicated warnings and sign-in-only info lines, group the rest by cause, and decide the order of repair.