Follow Discord
Sweep 30 Oct 2026 · 21:36Z Build v2.1.315 533 read Stable v2.1.308 Latest v2.1.314 Next v2.1.315 Feeds RSS JSON llms.txt llms-full.txt Unofficial

Claude Code v2.1.315 ·

dontAsk denials can note a tool could be approved, and are settled like background-agent denials

In dontAsk mode a refused tool call can now say it could have been approved, and is wrapped up the same way as background-agent refusals

Group of 2 You'll notice Improvements
JSON All of v2.1.315
You'll noticeTier: how much it should matter to you
2Useful: my rating, 1 to 5
2Signal: worth watching, 1 to 5
PermissionsArea: what it touches
ImprovementsKind: in v2.1.315,
ImprovementsSection of the release

What

dontAsk is a permission mode in which Claude Code does not show permission prompts. Actions that would normally need your approval are refused instead. The way these refusals are recorded and followed up has changed:

  • When a tool would have asked for permission in dontAsk mode, the refusal can now include an "ask" hint if the tool could have been approved. Before, it only recorded that the dontAsk mode caused the refusal.
  • A step settles a question that was never put to you. It used to run only for refusals from background agents, which are agents working on their own without you watching. It now also runs for refusals caused by dontAsk mode.
  • Sometimes a background agent's refusal overrides an approval from a hook. A hook is a script you configure that can approve or block actions. In that case the refusal now records hookAllowIgnored: true. That fact is passed to the settling step, together with whether the refusal came from a mode.

Why

A program hosting Claude Code gets more detail about an action refused in dontAsk mode, such as whether it could have been approved. That refusal now also goes through the step that settles the unasked question, where before this step was skipped for it.

Read from
Names in the bundledontAsk

See this entry in the whole of v2.1.315 →

Feedback