What happens to a Claude Code turn when the API times out mid-response?

In Claude Code 2.1.288, a mid-response API timeout no longer fails the turn for non-interactive sessions and subagents; they continue from the partial response instead. Thinking-only responses are retried. Unattended sessions using CLAUDE_CODE_RETRY_WATCHDOG now stream again after a failed long response and give up after three timeouts rather than retrying for hours.

How do I get more findings from /code-review in Claude Code?

Claude Code 2.1.288 adds a --max-findings option to /code-review. Pass a number, or all, to report more or fewer findings than the usual limit. The choice is remembered for later reviews until you pass --max-findings default, which returns /code-review to its standard number of findings.

Can I recover a prompt I cleared with Ctrl+C in Claude Code?

Yes. Claude Code 2.1.288 adds recovery for a prompt cleared with Ctrl+C: press the Up arrow on the empty prompt and the draft comes back. Anthropic's release notes say the recovered draft includes pasted text and images, not only typed text.

What security fixes are in Claude Code 2.1.288?

Claude Code 2.1.288 fixes a dangerous rm inside a bash -c or sh -c script running without a prompt in bypassPermissions mode or under a shell allow rule. It also prompts before a BASHPID assignment the shell would evaluate as arithmetic, and blocks a tool call when PreToolUse or PermissionRequest hook matching fails instead of skipping the hook.

Anthropic · 2026-10-02 · major

Claude Code 2.1.288 — timed-out replies continue instead of failing

Claude Code 2.1.288 lets headless sessions and subagents continue from a partial reply after an API timeout. It also brings back a prompt cleared with Ctrl+C, adds --max-findings to /code-review, and fixes many resume and plugin bugs.

Claude Code stops losing whole turns to API timeouts and gets back drafts you clear by mistake.

Quick facts

Version	v2.1.288
Maker	Anthropic
Released	2026-10-02
Reliability	Mid-response API timeouts resume from the partial reply
New flag	/code-review --max-findings <n>\|all
Rename	claude project purge → claude purge

What is it?

A mid-response API timeout no longer fails the turn in Claude Code 2.1.288: non-interactive sessions and subagents continue from the partial response, and thinking-only responses are retried. The release also lets you press Up on an empty prompt to recover a draft cleared with Ctrl+C, including pasted text and images.

How does it work?

Long conversations that reported zero token usage now auto-compact instead of failing with "Prompt is too long", and several --resume bugs that dropped files, thinking or the last answer are fixed. MCP servers that need more OAuth scope during a tool call now trigger a re-authenticate prompt, and mods get a $.ui.selection() call that returns the text you last selected in fullscreen mode.

Why does it matter?

Unattended runs in CI and the Agent SDK are where a single timeout used to waste the most work, so continuing from the partial reply saves both tokens and time. Teams that run /code-review can now ask for more or fewer findings than the default, and plugin authors get fixes for LSP placeholders, worktree subagents and older git versions.

Who is it for?

Claude Code users, CI and Agent SDK users, plugin authors

Frequently asked questions

What happens to a Claude Code turn when the API times out mid-response?: In Claude Code 2.1.288, a mid-response API timeout no longer fails the turn for non-interactive sessions and subagents; they continue from the partial response instead. Thinking-only responses are retried. Unattended sessions using CLAUDE_CODE_RETRY_WATCHDOG now stream again after a failed long response and give up after three timeouts rather than retrying for hours.
How do I get more findings from /code-review in Claude Code?: Claude Code 2.1.288 adds a --max-findings option to /code-review. Pass a number, or all, to report more or fewer findings than the usual limit. The choice is remembered for later reviews until you pass --max-findings default, which returns /code-review to its standard number of findings.
Can I recover a prompt I cleared with Ctrl+C in Claude Code?: Yes. Claude Code 2.1.288 adds recovery for a prompt cleared with Ctrl+C: press the Up arrow on the empty prompt and the draft comes back. Anthropic's release notes say the recovered draft includes pasted text and images, not only typed text.
What security fixes are in Claude Code 2.1.288?: Claude Code 2.1.288 fixes a dangerous rm inside a bash -c or sh -c script running without a prompt in bypassPermissions mode or under a shell allow rule. It also prompts before a BASHPID assignment the shell would evaluate as arithmetic, and blocks a tool call when PreToolUse or PermissionRequest hook matching fails instead of skipping the hook.

Try it

npm i -g @anthropic-ai/claude-code@2.1.288