Claude Code v2.1.183 blocks destructive git and infra commands in auto mode
Claude Code v2.1.183 tightens auto-mode safety by blocking destructive git commands and infrastructure teardown operations unless explicitly requested, alongside a broad set of bug fixes.
Score breakdown
The auto-mode safety guardrails directly prevent the agent from executing irreversible git and infrastructure teardown operations without explicit user intent, reducing the risk of accidental data or state loss during autonomous sessions.
- 01Destructive git commands (`git reset --hard`, `git checkout -- .`, `git clean -fd`, `git stash drop`) are now blocked in auto mode unless the user asked to discard local work
- 02`git commit --amend` is blocked when the commit was not made by the agent in the current session
- 03`terraform destroy`, `pulumi destroy`, and `cdk destroy` are blocked unless the specific stack was requested
Claude Code `v2.1.183` ships a set of auto-mode safety improvements designed to prevent the agent from silently executing destructive operations. Specifically, `git reset --hard`, `git checkout -- .`, `git clean -fd`, and `git stash drop` are now blocked unless the user explicitly asked to discard local work; `git commit --amend` is blocked when the commit was not made by the agent in the current session; and `terraform destroy`, `pulumi destroy`, and `cdk destroy` are blocked unless the user asked for the specific stack. The release also adds a deprecation warning for requested or frontmatter-configured models that have been deprecated or auto-updated, surfaced on stderr in print mode (`-p`).
The `/config` toggle behavior was also updated so that both Enter and Space change the selected setting, while Esc now saves and closes rather than reverting.
New configuration options include `attribution.sessionUrl` to omit the claude.ai session link from commits and PRs in web and Remote Control sessions, and `/config --help` to list all available shorthand keys for `/config key=value`. The `/config` toggle behavior was also updated so that both Enter and Space change the selected setting, while Esc now saves and closes rather than reverting. The startup "setup issues" line under the logo has been removed in favor of `/doctor` or `--debug`.
The bug fix list is extensive: it resolves `thinking.disabled.display: Extra inputs are not permitted` 400 errors on subagent spawns, empty WebSearch results in subagents, terminal cursor stranding in vim mode with the native cursor enabled, fullscreen TUI corruption (statusline mid-screen, duplicated spinner rows, merged text) in Windows Terminal under heavy nested-subagent load, silent turn completion when the model returned only a thinking block (Claude now re-prompts once), duplicate user-level skills in slash-command autocomplete with multiple plugins, MCP servers exposing auth-stub tools to the model in headless/SDK mode, tmux teammate pane launch failures with slow shell rc-file initialization, background tasks being killed when a teammate finishes a turn, scheduled task and webhook triggers being treated as keyboard input (they now classify as task notifications and can no longer approve pending actions or set the session title in auto mode), and focus mode incorrectly showing hook timing lines under each response.
Key facts
- 01Destructive git commands (`git reset --hard`, `git checkout -- .`, `git clean -fd`, `git stash drop`) are now blocked in auto mode unless the user asked to discard local work
- 02`git commit --amend` is blocked when the commit was not made by the agent in the current session
- 03`terraform destroy`, `pulumi destroy`, and `cdk destroy` are blocked unless the specific stack was requested
- 04New `attribution.sessionUrl` setting omits the claude.ai session link from commits and PRs in web and Remote Control sessions
- 05`/config --help` added to list all available shorthand keys for `/config key=value`
- 06Scheduled task and webhook trigger deliveries are now classified as task notifications and can no longer approve a pending action or set the session title in auto mode
- 07Fixed fullscreen TUI corruption in Windows Terminal under heavy nested-subagent load
Topics
Summary and scoring are generated automatically from the original article. We always link back to the publisher and never republish images or paywalled content. Last processed Jun 19, 2026 · 10:25 UTC. How this works →