Follow Discord
Sweep 06 Nov 2026 · 01:32Z Build v2.1.319 537 read Stable v2.1.311 Latest v2.1.318 Next v2.1.319 Feeds RSS JSON llms.txt llms-full.txt Unofficial

Claude Code v2.1.319 ·

Clearer retry guidance for the model when auto mode's safety check gives no answer

When auto mode's safety check returns no verdict, the model is now told either to retry once or that retrying now will not help, depending on the case

Under the hood Improvements
JSON All of v2.1.319
Under the hoodTier: how much it should matter to you
1Useful: my rating, 1 to 5
1Signal: worth watching, 1 to 5
Auto ModeArea: what it touches
ImprovementsKind: in v2.1.319,
ImprovementsSection of the release
What

In auto mode, Claude Code asks a server-side safety check whether each action is allowed. Sometimes that check returns no verdict. The text the model sees in that case used to be one fixed message: "a later call may get a verdict. Don't retry it now."

That text is now split by case:

  • One version tells the model to issue the call again once.
  • The other says that retrying now will not get a verdict.
  • When the server check was skipped, the model is told the call did not run yet, that this is a routine re-check and not an error or a refusal, that it says nothing about whether the action is safe, and that it should reissue the call once in its next response.

A denial caused by there being no check in the session now also carries the code auto_mode_no_check_in_session. The log line written when the check is unavailable for the AskUserQuestion tool, which then falls back to showing you the question dialog, now names that tool specifically. Claude Code's usage reporting also gains several new fields.

Why

The model gets more accurate advice about whether to try again when the safety check returns nothing, so it is less likely to give up on an action that would have worked or to keep retrying one that will not.

See this entry in the whole of v2.1.319 →

Feedback