You'll noticeTier: how much it should matter to you
3Useful: my rating, 1 to 5
2Signal: worth watching, 1 to 5
SandboxArea: what it touches
ImprovementsKind: in v2.1.323,
ImprovementsSection of the release
What
The sandbox is a restricted environment that limits what commands Claude runs can reach. If the settings you pass with --settings, or through the SDK, configure the sandbox but fail validation, none of them apply and Claude Code reports an error when it starts. This sits behind a server-side switch that is on unless it is switched off remotely.
Separately, if Claude Code cannot read its settings, the sandbox now stays on, and its automatic approval is turned off until the settings can be read. Before, whether the sandbox was on followed the settings alone.
Why
A mistake in sandbox settings no longer leaves you running with less protection than you asked for. Check startup errors if Claude Code refuses your settings.
Nothing here resolved what this flag was doing on this version, so nothing here should be read as on or off.
This account: no value returned · anonymous baseline: no value returned · compiled default in v2.1.323: on
These values were read against a different version of Claude Code, so treat them as the nearest reading available instead of one taken on this release.
Two sources agreeTwo things we can check say the same as this entry.
Anthropic's release notes agreeFixed claude -p and the Agent SDK running Bash unsandboxed when --settings or SDK settings that turn the sandbox on fail validation; Claude…
The name it cites is new in this buildNew in this build: tengu_mossy_karp
