Claude Code managed settings and admin controls for companies

Managed settings are the Claude Code rules a company sets once and applies to everyone. Anthropic describes them as "the settings your organization deploys to every developer's machine," applied "above every other level, so no user, project, local, or --settings value overrides them," apart from a few security-sensitive settings where a stricter lower-level value still counts. They are how an admin makes Claude Code refuse bypass mode, keep its file tools away from the production .env file, require sign-in to the company account rather than a personal one, and allow only approved tool connections.

There are two ways to deliver them. Server-managed settings are configured in the claude.ai admin console on Team and Enterprise plans and fetched by Claude Code automatically. Endpoint-managed settings are a file or operating-system policy placed on each machine, usually through device management software. This guide covers both as of October 2026, what a website company should enforce, how to confirm it worked, and the limits Anthropic itself states.

When a company needs managed settings

A single owner editing their own site can rely on project settings in .claude/settings.json, committed to the repository. Managed settings start to matter when any of these is true: several people or contractors use Claude Code on company code; you have client or regulated data; you need to prove what was allowed; or you want the same rules on every repository without trusting each one's settings file. The project-level baseline those rules build on is in whether Claude Code is safe for a live website.

They sit on top of the settings precedence order, from highest to lowest: managed policy, then --settings, then .claude/settings.local.json, then .claude/settings.json, then ~/.claude/settings.json. Account-level controls are separate: Anthropic notes that "SSO, SCIM provisioning, and seat assignment are configured at the Claude account level." SSO is available on Team and Enterprise, and SCIM on Enterprise only.

Choose a delivery mechanism

Anthropic's admin setup guide lists four sources, checked in priority order:

Two rules about combining them. When more than one source reaches a machine, "Claude Code by default uses one and ignores the others." And endpoint-managed settings "don't reach cloud sessions in Anthropic-hosted environments," so if your people use Claude Code on the web, configure server-managed settings as well. The admin guide says the HKCU registry is writable without elevation, so treat it "as a convenience default rather than an enforcement channel."

Deploy a managed settings file, step by step

A managed settings file uses the same JSON shape as a normal settings.json. Anthropic's starter example blocks two file reads, turns off bypass mode and makes managed settings the only source of permission rules:

{
"permissions": {
"deny": ["Read(./.env)", "Read(./secrets/**)"],
"disableBypassPermissionsMode": "disable"
},
"allowManagedPermissionRulesOnly": true
}

1. Write the file with the keys you have decided to enforce, starting from an example like the one above.
2. Place it on each machine at the system path: macOS /Library/Application Support/ClaudeCode/managed-settings.json; Linux and WSL /etc/claude-code/managed-settings.json; Windows C:\Program Files\ClaudeCode\managed-settings.json.
3. Check one machine first. Run /status inside Claude Code. The Setting sources line should read "Enterprise managed settings (file)."
4. Validate the JSON. Anthropic warns that if a managed settings file "isn't valid JSON," Claude Code "refuses to start," which is safe but will stop everyone at once.
5. Roll out to the fleet, then spot-check with /status. If a rule seems ignored, claude doctor lists entries Claude Code dropped.

For server-managed settings the steps are the same idea: an Owner edits the JSON in the admin console, and machines pick it up at startup and hourly after that. Certain settings, including hooks and settings that run shell commands, trigger an approval dialog for each user, and "If a user rejects the settings, Claude Code exits." The full details are on Anthropic's server-managed settings page.

What a website company should enforce

Anthropic's admin guide groups the controls by area. A reasonable baseline for an agency or an in-house website team:

A few notes on these. Permission rules and the sandbox cover different layers: "Denying WebFetch blocks Claude's fetch tool, but if Bash is allowed, curl and wget can still reach any URL. Sandboxing closes that gap." The mode decision is explained in every permission mode explained. Plugin governance is covered in plugins and marketplaces, and hooks in hooks for website projects. A managed model value "is a default, not a lock"; use availableModels to restrict choices.

Sign-in and account controls

A policy only helps if people are signed in to the company's organization. forceLoginMethod and forceLoginOrgUUID restrict login to a method and to your organization, and when they are set, sessions that authenticate with an environment API key are blocked at startup. Anthropic's authentication docs add an important catch: server-managed settings "reach only accounts that are already authenticated into your organization, so they can't redirect a developer's first login." Deploy the login keys through device management or a settings file.

Outside managed settings, Team and Enterprise Owners also control organization-wide features in the admin console, such as whether Remote Control and cloud sessions are allowed. Spend limits can be set at the organization, group or individual member level, and Team and Enterprise have an analytics dashboard for usage.

The limits Anthropic states

Managed settings are strong, but they are not magic, and the docs are candid about it:

• Server-managed settings are client-side. "On unmanaged devices, a user doesn't need admin or sudo access to bypass them." A user who switches to a third-party provider such as Bedrock, or signs in to a different organization, doesn't receive them.
• Local admins can edit local policy. A developer who is an administrator on the machine "can edit the managed source itself," which is why device management tools redeploy files on a schedule.
• One policy for everyone. Server-managed settings "apply uniformly to all users in the organization. Per-group configurations are not yet supported." Different groups need different files or profiles, or a self-hosted Claude apps gateway, which can deliver settings per identity-provider group.
• Claude Code only. Other tools that call the API aren't bound by them.

For stricter guarantees, Anthropic recommends endpoint-managed settings on devices enrolled in device management, and an option called forceRemoteSettingsRefresh that makes Claude Code exit if it can't fetch server-managed settings at startup.

Data and audit settings

Managed settings also carry data rules. Through the env block you can set DISABLE_TELEMETRY for everyone, though Anthropic notes this also stops the usage data that feeds your analytics dashboard. cleanupPeriodDays sets how long every machine keeps local transcripts. Audit events for settings changes are available "through the compliance API or audit log export," via your Anthropic account team. How retention and training work by plan is covered in what happens to your code and data.

Roll new rules out to a pilot group first, as described in our rollout plan for teams, and keep a copy of the policy in version control so changes are reviewed like code. Copy-ready project settings files are on our Claude Code templates page.

Conclusion

Managed settings turn a team's Claude Code rules from a shared document into enforced policy: bypass mode off, secrets denied, approved tools only, and sign-in tied to the company account. Pick the delivery mechanism that matches how you manage devices, add server-managed settings if people use cloud sessions, verify with /status on real machines, and remember Anthropic's own caution that a client-side control is not a security boundary on unmanaged laptops. If you want a policy designed, piloted and documented for your team, that is part of our Claude Code consulting. More guides are in our Claude Code Guide.

Frequently asked questions