You'll noticeTier: how much it should matter to you
4Useful: my rating, 1 to 5
2Signal: worth watching, 1 to 5
Managed SettingsArea: what it touches
ImprovementsKind: in v2.1.300,
What probably matters to youSection of the release
Unclear It is not confirmed that Claude Code acts on these settings yet, only that they now list it as a reader.
What
Some settings live in a shared, centrally managed configuration that several Anthropic apps read. Each setting lists its readers, such as "desktop" or "m365". Many now also list "code", meaning Claude Code:
Gateway and inference: inferenceGatewayBaseUrl, inferenceStreamIdleTimeoutSec, isLocalDevMcpEnabled, inferenceCustomHeaders, inferenceFoundryResource, and the inference credential helper. The credential helper also lost its executionTarget: "desktop" restriction.
Other: endUserAttribution, scheduledTasksEnabled and builtinToolPolicy.
Before, these were read only by "desktop", or by "desktop" and "m365". A new activatesThirdPartyMode export was also added. A new message reads "To sandbox sessions, set sandbox.enabled to true in settings.", referring to sandbox.enabled. For the OTLP and attribution settings, the change seen is in the schema only.
Why
Organisations that manage Desktop through this configuration, for example to route requests through their own gateway or cloud provider or to export telemetry, can drive Claude Code from the same settings.
How sure we are
Something disagreesSomething we can check disagrees with this entry, or the writer said they could not settle it.
The writer flagged doubtIt is not confirmed that Claude Code acts on these settings yet, only that they now list it as a reader.
Anthropic's release notes agreeRemoved the --sandbox and --no-sandbox flags from claude remote-control, which had no effect; passing either now exits with an error. Set…
