Workbranch › Guides

How to resume a Claude Code session after a break or the usage limit

You closed the terminal, your laptop slept, or Claude Code stopped at the usage limit halfway through a task. The conversation isn't gone. This guide shows how to get back into it, how to wait out a limit, and how to make any new session start with the context it needs.

Updated 20 October 2026

Commands and behavior below were checked against Anthropic's Claude Code documentation as of October 2026. Claude Code changes often, so run claude update if something described here is missing.

Resume the last session in a folder

Claude Code saves every session to a local transcript as you work, tied to the folder you started it in.

1. Open a terminal and cd into the same project folder.
2. Run claude --continue. The most recent conversation from that folder opens with its history.
3. Type your next instruction, for example carry on with the migration from where you stopped.

A resumed session keeps its messages and tool results, and continues on the model it was using. A tool that was still running when the old process ended is marked as cut off, and Claude is told to check whether it took effect before running it again. Flags you passed at launch, such as --mcp-config or --add-dir, aren't restored, so pass them again.

Pick an older session

1. Run claude --resume (or claude -r) with no arguments to open the session picker. Inside a running session, /resume does the same.
2. Move with the arrow keys, press Space to preview a session, and start typing to search.
3. Press Ctrl+A to include sessions from every project on the machine, or Ctrl+B to show only the current git branch.
4. Press Enter to resume the highlighted session.

To skip the picker next time, name your sessions. Start with claude -n auth-refactor, or run /rename auth-refactor mid-session, and come back with claude --resume auth-refactor.

When the usage limit stopped Claude

Claude Code shares your claude.ai plan's five-hour session limit and weekly limit. What happens at the limit:

• Wrapping up. If the limit hits mid-response, Claude Code may show Usage limit reached · wrapping up and finish the current step. Anthropic is rolling this out gradually.
• Automatic continue. From version 2.1.234, in an interactive session signed in with a claude.ai subscription, Claude Code shows when your limit resets and that it will continue then. Keep the terminal open and it picks the task up after the reset. Press Esc at an empty prompt to cancel.
• Not for weekly limits. Claude Code doesn't start this wait on its own when the reset is more than 24 hours away, which a weekly limit can be. It offers a usage-limit options menu instead.
• After sleep. If your computer slept for more than about half an hour and the limit reset meanwhile, you'll see a prompt to press Enter to continue.
• If you exited. The wait doesn't restart when you resume. Run claude --continue after the reset and ask Claude to carry on.

Run /usage to see your plan limits and what is counting against them. Weekly limits can be days away; for the options in that case, see Claude usage limit reached: what to do.

Clean up context: /compact or /clear

A resumed session brings back everything, including a context window that may already be crowded. Two commands help:

Claude Code also compacts automatically as the window fills. Run /context to see what is using space.

Give every new session the context with CLAUDE.md

A resumed session remembers the conversation; a new one starts from scratch. A CLAUDE.md file in the project root closes that gap, because Claude Code loads it at the start of every session and re-reads it from disk after compaction.

1. Run /init in the project. Claude looks at the codebase and writes a starter CLAUDE.md with build commands, tests and conventions.
2. Edit it with /memory or in your editor. Add what Claude can't work out from the code, like project decisions and environment quirks.
3. Keep it short. Anthropic suggests under 200 lines; long files cost context and get followed less reliably.
4. Run /context in a new session to confirm the file loaded.

Instructions you only typed in chat may be lost when a session is compacted. If one matters, move it into CLAUDE.md.

If it doesn't work

• --continue opens the wrong session or none. It only looks at the current folder. Check you're in the right directory, or use claude --resume and widen with Ctrl+A.
• A session run with claude -p is missing. Non-interactive and Agent SDK sessions are left out of the picker and --continue. Resume one with its session ID.
• No conversation found with session ID. The ID is wrong, or the transcript was cleaned up after the retention period.
• Claude forgot a rule after resuming. The rule was probably only in the conversation and got summarized away. Put it in CLAUDE.md.

See what's still open when a session starts

Transcripts and CLAUDE.md hold the code side, but not always the open to-dos and decisions. The Workbranch plugin for Claude Code adds that: run /workbranch:save before you stop, and Claude files the session's tasks, decisions and open questions into a Workbranch project. Put the project's name in a .workbranch file at the folder's root, and every new session there opens with a recap of what's still open. Workbranch has a 3-day free trial at app.workbranch.ai.

Sources

• Manage sessions (Claude Code Docs, checked 2026-10-20)
• CLI reference (Claude Code Docs, checked 2026-10-20)
• Commands (Claude Code Docs, checked 2026-10-20)
• Interactive mode (Claude Code Docs, checked 2026-10-20)
• Explore the context window (Claude Code Docs, checked 2026-10-20)
• How Claude remembers your project (Claude Code Docs, checked 2026-10-20)
• Best practices for Claude Code (Claude Code Docs, checked 2026-10-20)
• Use Claude Code with your Pro or Max plan (Claude Help Center, checked 2026-10-20)

Frequently asked questions