Service eligibility and regional restrictions
PuppyIP serves only compliant overseas businesses and their authorized personnel. Proxy services are not available in mainland China. The service may only be used for lawful business activities outside mainland China. Use of this service within mainland China is prohibited.
Hosting a proxy IP or server overseas does not change these restrictions. The service must not be provided to end users in mainland China through relaying, forwarding, sharing or resale. Before use, read the Terms of Service.
Key Takeaways
- Prompt hooks and pre-tool-call hooks are generally available; several other security and admin features remain in Beta.
- Inference hooks send prompts or tool calls to your security service and wait for an allow or deny decision before Claude continues.
- Centralized parent-organization settings require approval from administrators of associated organizations; the final approval completes an irreversible setup.
- Member sign-out does not revoke API keys or end running cloud sessions and scheduled routines; EDR detects activity in a local VM but does not block it or cover cloud Cowork.
Separate GA features from Beta
On November 5, 2026, Anthropic announced security and administration updates for Claude Enterprise, including a security center, inference hooks, centralized management, and new access controls. The release notes and product article state that prompt hooks and pre-tool-call hooks are generally available. The security center, centralized management, Admin API member sign-out, self-serve IP allowlisting, Claude Desktop admin settings, and EDR are each still in Beta.
Enterprise Owners can first list the Claude, Claude Code, Cowork, and Desktop entry points their teams use, then check each required feature against its plan, role, and runtime scope. The reviewed official sources do not give a complete feature-specific regional list, separate pricing, or an activation time for every account. One feature’s eligibility does not establish universal access to the entire announcement.
Inference hooks: check prompts and tool calls against policy
Inference hooks connect Claude to your organization's existing data-security tools. A prompt hook checks a prompt before Claude starts generating a response. A pre-tool-call hook checks a tool call before the tool runs. Your organization-hosted service returns allow or deny under your policy, and Claude continues after it receives that decision. Both hook types are generally available and can be enabled and managed by Enterprise Owners or Primary Owners.
One organization-level setup can cover Claude, Claude Code, Cowork, and other Claude Enterprise products, including tool calls made through skills, plugins, and connected tools. Checks run inside Claude’s infrastructure without software installed on employee devices; your organization-hosted service makes the policy decision. For implementation, follow the official technical documentation for endpoint configuration, request-signature verification, and monitoring.
Security center: review guidance, not a compliance certification
The security center is still in Beta. It is a central page in Organization settings for reviewing security settings for Claude chat, Cowork, and Claude Code against Anthropic's recommended configuration. It shows current values, recommended values where available, and links to the relevant settings. It also identifies controls that are new or recently changed, and rechecks automatically checked settings whenever the page loads.
Admins can work together to record each control's status, accept risks, or flag issues, then download a report for audit discussions. The report distinguishes automatically checked controls from those confirmed by an Owner. Organizations with the Compliance API can read the same results. This is security guidance based on Anthropic's recommendations, not a certification or guarantee of compliance with any law or standard.
Centralized management requires cross-organization approval
Centralized management is still in Beta. It gives a parent organization representing the company's Claude Enterprise and Claude Platform organizations an admin console at admin.claude.com. Admins can manage company-wide settings there, including single sign-on, SCIM directory sync, domain verification, groups, and Compliance API keys. Each organization keeps its own members, roles, and organization-level controls such as IP allowlisting, tenant restrictions, and session duration.
An admin of any organization under the parent can propose the parent-organization admins. Admins of the other organizations review and approve; setup completes when the last approval is recorded. Setup cannot be undone, so identify the parent admins before starting. A parent role does not grant access inside each organization by default, although Compliance API keys created at the parent can read across organizations.
Sign-out, IP allowlisting, and Desktop settings have separate limits
The Admin API member sign-out feature is in Beta. One call ends that member's sign-in sessions on all browsers and devices, so apps such as Claude Code need to sign in again. It does not revoke API keys, end cloud sessions that are already running, or stop scheduled routines. The member can sign in again afterwards, so sign-out is not account deactivation or credential rotation.
Self-serve IP allowlisting is in public Beta for Claude Enterprise and Claude Platform organizations. Enterprise Owners and Claude Platform Admins can manage it themselves. It limits access to most Claude products to organization-approved IP addresses. Some Claude Desktop settings are also in Beta in Organization settings, including blocking personal and other outside Claude accounts on managed computers; Owners can also limit which workspace folders Claude can use.
EDR covers the local VM and runs in detection mode
When Cowork and unified Claude work locally on a user's computer, shell commands and code run inside an isolated virtual machine. The endpoint detection and response (EDR) Beta can place your organization's sensor in that local VM and report its activity to the console your security team monitors. It does not cover Cowork sessions running in the cloud.
The integration is deployed through device management on macOS and Windows. Its stated scope includes Team and Enterprise and the Claude Desktop app used through Amazon Bedrock, Google Cloud Vertex AI, or Microsoft Foundry, deployed through MDM for supported vendors. Claude verifies the sensor package signature each time the VM starts. Sensors run in detection mode: they report activity but do not block it.
Sources
Frequently Asked Questions
How do parent-organization Owner and Admin permissions differ?
Anthropic says parent Owners can manage all parent-level settings, including assignment of Owner and Admin roles. Parent Admins can manage the other parent-level settings but cannot assign those roles. Parent-level role assignment is separate from membership permissions inside each organization.
Can a company retain multiple identity providers after an acquisition?
Yes. Anthropic says centralized management supports multiple identity providers, each with its own single sign-on and SCIM configuration. An acquired company can retain its provider, or a team can migrate one domain at a time. This describes product support; it is not evidence that PuppyIP configured or migrated an organization.
Does IP allowlisting restrict Artifacts shared outside the organization?
No. Anthropic's IP allowlisting help page explicitly says that Artifacts shared outside the organization are not restricted by the allowlist.