Let Claude Code continue by itself after a limit resets

Leave the Claude Code session open. When a claude.ai usage limit stops Claude mid-task, Claude Code waits in that session and continues the task on its own after the limit resets. Anthropic's interactive mode docs call this automatic continue. It is on by default in interactive sessions signed in with a claude.ai subscription and needs Claude Code v2.1.234 or later.

What you see

While Claude Code waits, two lines at the bottom of the session show when the limit resets and when Claude will continue, for example "Usage limit reached · limit resets 3:45pm" and "Continuing automatically at 3:45pm · Esc to cancel". The times are examples from the docs.

At the reset, Claude Code tells Claude to pick the task up where it stopped. It does not resend your last message.

The important condition is that the session stays open. The docs are explicit: if you exit Claude Code, the wait ends, and it does not restart when you resume the session later.

When it starts on its own, and when it does not

Automatic continue is deliberately bounded. According to the docs:

• It stops after three automatic continuations, with the notification "Automatic continue stopped after repeated usage-limit hits". Sending another prompt starts the count again.
• It does not start on its own when the reset is more than 24 hours away. A weekly limit can be days out.
• It does not start on its own in Remote Control sessions or agent-team teammate sessions.
• It does not start on its own for an Opus or Sonnet limit while you are running a model outside that family, since your next turn may not hit that limit.

In those cases, a wait is still available if you want one. Claude Code opens its usage-limit options menu once per reset window when you hit a limit at your own terminal, and you can pick the row that starts "Wait here, then continue automatically". You can also open the menu yourself with /rate-limit-options.

It is not offered at all in background sessions, -p runs, or with API keys, cloud providers or usage-based billing, where usage is metered per request and there is no reset to wait for. See does claude -p use your subscription limits.

So in practice it is mostly a tool for the five-hour session limit. For a weekly limit, see what to do when you hit a weekly limit.

Cancelling and turning it off

To cancel a wait, press Esc at an empty prompt or Ctrl+C while the lines show, or pick "Don't continue automatically" in /rate-limit-options. A wait also ends without continuing if you send a prompt, clear or rewind the conversation, /resume another session, or switch accounts with /login.

To turn the feature off, open /config and turn off "Continue automatically at usage limit", or set the key in your user settings file (~/.claude/settings.json, or the settings.json in your CLAUDE_CONFIG_DIR folder):

{
  "autoContinueAtUsageLimit": false
}

The settings reference says this key is read from user settings, --settings and managed settings. With it off, Claude Code no longer starts a wait by itself, but you can still start one from the menu.

Turn it off if you would rather decide yourself, for example when the task is one you want to review before it runs again.

/rate-limit-options

The /rate-limit-options command shows ways to keep working when a claude.ai usage limit blocks a request: wait and continue automatically, add usage credits, or upgrade. It requires a claude.ai subscription, per the commands reference.

If you add usage credits with /usage-credits during a wait, Claude Code checks whether usage is available again and continues right away if it is. Usage credits are billed separately from your subscription, at standard API rates, and they do not change when your limits reset. They are explained in usage credits on Claude plans.

The Claude Desktop checkbox

In the Claude Desktop Code tab, the session-limit card has an "Auto-continue when limits reset" checkbox. The weekly-limit card does not. Per the error reference, this is separate from the CLI setting, so turning one off does not turn the other off.

What happens at the limit before it waits

Some plans get a short wrap-up allowance, where Claude Code may keep working briefly to finish the step in progress and shows "Usage limit reached · wrapping up". It is small and capped, counts toward your weekly limit, and does not apply to new messages sent after the limit. It is rolling out gradually and needs Claude Code 2.1.277 or later, per the Help Center. Do not rely on it being there.

Is it worth leaving running?

Automatic continue suits a task you have already scoped and trust: a long refactor, a test-fix loop, a migration with clear checks. It suits a task badly where you wanted to review each stage, because the work resumes without you. It also resumes with exactly the instruction you left it with. If that instruction was vague, waiting will not improve it: cancel, give a sharper instruction, and continue yourself.

Why a job did not continue

If you expected a task to resume overnight and it did not, check this list:

1. Was the session still open, and did you leave it alone? Exiting, sending a prompt or clearing ends the wait.
2. Was the limit a weekly one with a reset more than 24 hours away?
3. Had it already continued three times?
4. Were you on an API key, cloud provider or usage-based billing?
5. Was the limit for one model family while you ran a different one?
6. Is it waiting on a permission prompt? The docs note that a continued task still asks for permissions as usual, so it can stop on a prompt while you are away.
7. Did the Mac sleep? If it slept for more than about 30 minutes and the limit reset meanwhile, Claude Code shows "Press enter to continue" and waits for you.
8. Did the login expire in the meantime? Users have reported a login expiring and stopping overnight work (GitHub issue 104806). Anthropic's authentication docs describe a warning at startup, "Your login expires in 3 days · run /login to renew", and a later "Login expired · Please run /login" error. See Claude Code login expired.

Know the reset time yourself

Automatic continue relies on the reset time shown in the message. To know when each of your accounts comes back without opening a session, Headroom, a free menu bar app that is independent and not made or endorsed by Anthropic, lists each Claude Code account signed in on your Mac. An account whose limit is used up shows a status such as "Back in 38m", and opening its row shows each limit with a countdown and clock time, such as "Resets in 38m, 3:40 PM". It never sends a prompt, so checking does not use your limits. Download Headroom.