Claude Code v2.1.234 now resumes your session automatically when your usage limit resets

Hey there, it's me! I just caught the latest Claude Code release, so let's dive right in.

Claude Code Release github.com

What was announced?

Over on Claude Code Release, the notes for version v2.1.234 went live. It adds automatic session resumption when your claude.ai usage limit resets and stronger GitLab integration, alongside Windows security hardening and a long list of reliability fixes.

The story so far

Until now, if you hit your claude.ai usage limit and your session stopped, you had to send a signal yourself to resume once the limit reset. GitLab merge request info showed up in the --worktree flag and the claude agents view, but there was no badge for it in the footer or status line. Hosts that give each session its own config directory had no way to pick a short name for the per-project transcript directory. And on Windows, paths written with the NT namespace (\??\) prefix had UNC path validation partially closed in v2.1.233, but remote file reads, session restore, CLAUDE.md includes, workflow scripts, and file uploads still hadn't been hardened against it.

What changes

This release adds several things you can put to use right away:

• Claude Code now automatically resumes your session once your claude.ai usage limit resets (turn it off in /config under "Continue automatically at usage limit")
• A new GitLab merge request badge in the footer and status line: repos with a GitLab remote and an authenticated glab CLI show MR !N with draft/pending/green states
• A new CLAUDE_CODE_PROJECT_DIR_NAME environment variable lets hosts that give each session its own config directory choose a short name for the per-project transcript directory
• A new selection:clear keybinding action lets you bind a key to clear an in-app text selection (works in the agents view too)
• Claude is now told to use your account email only to identify you, not to send it to unrelated services unless you ask
• The Windows NT-namespace (\??\) path hardening is extended: remote file reads, session restore, CLAUDE.md includes, workflow scripts, and file uploads now all reject these paths, closing more of the NTLM credential-leak vector
• Fixed auto mode in very long sessions repeatedly re-checking and denying sandboxed commands' network access after the conversation had been compacted
• Fixed session-scoped permission answers (including denies) being dropped when answering background subagent tool permission prompts
• Fixed a crash when an API response on the non-streaming fallback path (typically via third-party gateways) contained a thinking block missing its thinking field or a text block missing its text field

Dive Deep

There's a long tail of smaller fixes too — here are the ones most likely to matter day to day.

• Lighter built-in skill: the context cost of loading the claude-api skill dropped from over 200k tokens to about 25k, by loading reference docs on demand instead of all at once
• Config changes mid-turn: /permissions can now be opened while Claude is working, with rule changes applying to the rest of the current turn. /add-dir {path} works mid-turn too, and in the fullscreen TUI, /add-dir, /autocompact, /theme, /help, /config, and /advisor dialogs can all open mid-turn
• Smarter /goal: when a turn dies on an unrecoverable error (revoked auth, an exhausted credit balance, a context overflow), /goal now clears itself with a notice instead of staying armed. When background tasks keep a goal waiting 30+ minutes, Claude now checks in on them (set CLAUDE_CODE_GOAL_CHECKIN_MINUTES=0 to opt out)
• Better session titles: auto-generated session titles now read as short, specific names (like "Login button bug") instead of long sentences restating your request (like "Fix the login button on mobile")
• More reliable Remote Control: signing this computer in to a different claude.ai account or organization now stops the running session within seconds and explains why, instead of a confusing HTTP 404 hours later. Phones and claude.ai/code also stay in sync with permission mode, model, and effort changes as they happen
• Quiet privacy and security fixes: MCP diagnostics no longer print resolved secrets, and permission-preview credential masking can no longer be bypassed to leak commands, paths, or destinations — several small information-leak paths closed in one release
• Small UX polish: consecutive shell commands no longer split into separate rows when todo updates land in between, a queued shell command edited with the up arrow no longer gets sent as plain text, and a horizontal rule no longer crowds the line right after it

Wrap-up

• Sessions now resume automatically once your claude.ai usage limit resets, and a GitLab MR badge shows up in the footer/status line
• Windows NT-namespace path hardening now covers remote file reads, file uploads, and more
• The claude-api skill's load cost is down about 8x, from over 200k tokens to roughly 25k
• Several reliability fixes: auto mode's network permission re-checks, dropped subagent permission answers, and a non-streaming crash
• /goal, /permissions, and /add-dir all got easier to use mid-turn, among other small improvements

This one's worth a look if you use Claude Code daily — especially if you've ever been annoyed waiting out a claude.ai usage limit, if you work in a GitLab shop, or if you drive Remote Control sessions from your phone!