Agent Upstart

Cool things people are building with agents — and practical help getting the most out of them.

Playbooks · Teams tasks

Verify what managed settings auto-apply after 2.1.316

Written November 5, 2026

Server-managed settings apply without an approval prompt for Team and Enterprise admins. The docs and release notes disagree; check your machine.

The two wordings

Server-managed settings are the policy an Owner or Primary Owner sets in the claude.ai admin console for a Team or Enterprise organization. Claude Code fetches them from Anthropic's servers at startup and polls for updates hourly. The 2.1.316 release notes, published October 31, 2026, say the delivered settings now apply without an approval prompt, and that a stored copy that does not verify is refused.

The server-managed settings docs still describe the approval dialog. Their delivery check says that if the configuration includes settings that trigger the security approval dialog, the user sees a prompt describing the managed settings the next time Claude Code fetches them — at the next start, or within an hour in a running interactive session. The fetch section says the same thing from the other side: when the payload needs security approval, Claude Code ends its wait and applies the payload once the developer approves. The managed settings docs agree: a server-managed change to a setting that needs approval, such as a hook or an env variable, waits for the developer to accept the dialog in an interactive session.

Neither wording is picked here, because reading alone cannot settle which one your machine follows. The release notes say the prompt is gone; the docs say it is still there. The disagreement is the point of this page, and the check below settles it per machine.

What the dialog used to catch

Most managed policy never needed approval. The dialog covered the risky subset: settings that execute shell commands, such as apiKeyHelper, statusLine, and otelHeadersHelper; sandbox settings that point at an executable or weaken isolation; delivered env variables such as proxy and base-URL variables; and any hook definition. Team admins used that dialog as the moment they saw a new hook, helper, or proxy variable arrive. Under the release-notes wording, that moment is gone: the settings arrive and apply, and the first sign of a new hook is the hook running.

Three things do not change either way. The old dialog was never a veto — turning the settings down ended Claude Code rather than letting the session continue without them. Sessions running without interaction already applied approval-gated settings for that session only. And ConfigChange hooks do not run when server-managed settings arrive or refresh, so a hook cannot watch for its own replacement.

Check your machine

These steps run on your own machine against your own organization's settings. They show what your machine received; they do not change what the organization sends.

  1. Run claude doctor and read the Managed settings (remote) line. The docs say it reports one of four outcomes: the delivered settings loaded; your organization has no server-managed settings configured; the fetch failed, with the cause and whether a cached policy still applies; or Claude Code skipped the fetch, with the reason. The line needs Claude Code 2.1.248 or later. In a running session, /status shows the same line after a failed fetch.
  2. Read the cached settings file. The cache lives at ~/.claude/remote-settings.json. Read it and look for hooks, apiKeyHelper, statusLine, otelHeadersHelper, anything under sandbox, and the env block. What is in that file is what your machine is applying from the server — including any hook or helper that arrived without a dialog.
  3. Compare the two wordings against what you see. If a new hook, helper, or proxy variable appears in the cached file without a dialog having shown, your machine follows the release-notes wording. If a dialog still appears before approval-gated settings apply, it follows the docs. Either way, the cached file is the record of what actually arrived.
  4. Note what the cache holds back. Cached settings apply immediately at startup, except for some categories the docs say are withheld until the server confirms the payload — including proxy and TLS configuration, API routing variables, and credential-paired helpers. A value in the cached file that falls in a withheld category is not yet in effect, even though the file lists it.

Limits

Related playbooks

Sources

Back to: Small-team playbooks hub · Also: Claude Code 2.1.317: schedule warnings and cost reports · Claude Code 2.1.318: sign-in expiry and fork effort