Claude Code
- Auto mode now blocks destructive commands you didn't ask for.
git reset --hard,git checkout -- .,git clean -fd,git stash drop,git commit --amend(for commits the agent didn't make this session), andterraform destroy/pulumi destroy/cdk destroyare all gated (more below). - Esc in
/confignow saves and closes instead of reverting. If your muscle memory uses Esc to discard a toggle change, you'll commit it instead. - Scheduled tasks and webhook triggers can no longer approve pending actions in auto mode. They're reclassified as task notifications, so they can't greenlight a permission prompt or set the session title.
New in 2.1.183
2.1.183 (June 19, 2026)
- Improved auto mode safety: destructive git commands (
git reset --hard,git checkout -- .,git clean -fd,git stash drop) are now blocked when you didn't ask to discard local work,git commit --amendis blocked when the commit wasn't made by the agent this session, andterraform destroy/pulumi destroy/cdk destroyare blocked unless you asked for the specific stack - Added a warning when the requested model is deprecated or automatically updated to a newer model, shown on stderr in print mode (
-p) and now also covering models set in agent frontmatter - Added
attribution.sessionUrlsetting to omit the claude.ai session link from commits and PRs in web and Remote Control sessions - Added
/config --helpto list all available shorthand keys for/config key=value - Changed
/configtoggle behavior: Enter and Space both change the selected setting, and Esc now saves and closes instead of reverting - Removed the startup "setup issues" line under the logo, run
/doctorto see configuration issues or use--debug - Fixed
thinking.disabled.display: Extra inputs are not permitted400 errors on subagent spawns and session-title generation for affected configurations - Fixed WebSearch returning empty results in subagents
- Fixed the terminal cursor being stranded above the prompt after navigating history in vim mode with the native cursor enabled
- Fixed fullscreen TUI corruption (statusline mid-screen, duplicated spinner rows, merged text) in Windows Terminal under heavy nested-subagent load
- Fixed turns silently completing with no visible output when the model returned only a thinking block; Claude now re-prompts once
- Fixed user-level skills appearing multiple times in slash-command autocomplete when multiple plugins are enabled
- Fixed MCP servers requiring authentication exposing auth-stub tools to the model in headless/SDK mode
- Fixed tmux teammate panes failing to launch when the shell has slow rc-file initialization, and keystrokes typed during agent spawn leaking into the new tmux pane instead of the leader prompt
- Fixed background tasks started by a teammate being killed when the teammate finishes a turn
- Fixed scheduled task and webhook trigger deliveries being treated as keyboard input; they now classify as task notifications and can no longer approve a pending action or set the session title in auto mode
- Fixed focus mode showing "Ran N PostToolUse hooks" timing lines under each response
Notes
Auto mode safety keeps tightening
This is the third beat in a deliberate hardening arc. 2.1.178 added classifier evaluation before subagent spawns, closing the gap where a subagent could request a blocked action without review. 2.1.183 extends the net to the commands themselves: destructive git operations, amends of commits the agent didn't make, and infrastructure destroys are all blocked unless the user explicitly asked. If you run auto mode in CI or unattended workflows, these guardrails are now default with no opt-in needed. The --amend guard is particularly worth knowing: it prevents an agent from rewriting a commit it didn't author, which was a real failure mode (this repo learned that lesson on 2026-06-10).