MCP server record

  • Release version: Zurich
  • Updated November 23, 2025
  • 3 minutes to read
    • Summarize
    Summarized using AI
    This content was generated using new OpenAI-powered functionality. Results are provided on an as is basis and are not guaranteed to be accurate or complete.

    Summary of MCP server record

    The MCP server record in ServiceNow’s AI Control Tower provides a comprehensive interface to manage and monitor MCP servers, including their details, approval workflows, performance metrics, and AI Gateway configurations. This record plays a critical role in governing AI infrastructure by enabling approval processes, observability, and integration management to ensure secure and controlled AI operations.

    Show full answer Show less

    Tabs and Their Functions

    • Details: Displays all asset information including the MCP server URL for quick reference.
    • Approvals: Shows all approval requests related to the MCP server. Note that with the AI Control Tower Pro license, the approval button may not appear automatically after syncing MCP servers. A known workaround involves verifying lifecycle states and manually updating the governance record via a provided script to initiate approval workflows.
    • KPIs & Metrics: Provides observability metrics for approved and running MCP servers, showing total requests, success rates, and latency. AI stewards can monitor usage across hosts, with similar aggregated data available on the AI Gateway tab.
    • AI Gateway setup: Visible after MCP server approval, this tab allows pausing/resuming AI Gateway transactions per server and adding MCP Client integrations. It displays key URLs—Gateway endpoint, authorization, and token endpoints—all auto-generated and immutable.
    • MCP Client Integration: Facilitates registration of clients (e.g., agent builders) that require unique client IDs and secrets. Enables enabling AI Agent Studio for linear MCP servers by adding new client integrations.
    • Global MCP Clients: Managed separately on the AI Control Tower Configurations page, global clients apply across all MCP servers, eliminating the need for individual client setup per server.
    • Related Variants: Maps MCP server entries from AI Agent Studio to AI Control Tower for visibility and tracking. Supports multiple MCP servers of the same variant and includes options for activating sensitive data (PII) protection to block transactions sharing sensitive information.

    Practical Benefits for ServiceNow Customers

    • Centralized management of MCP servers with full visibility into asset details, approval status, and operational metrics.
    • Governance controls to initiate and manage approval workflows ensuring compliance before MCP servers are operational.
    • Observability tools to monitor server performance, usage, and latency, enabling proactive management and troubleshooting.
    • Secure AI Gateway integration setup with options to pause malicious servers and control client registrations, enhancing security and operational control.
    • Streamlined client integration processes for AI agent builders and global client sharing to simplify administration.
    • Enhanced data protection capabilities by enabling PII checks at the gateway level to prevent sensitive data exposure.

    Explore the MCP server record and its associated tabs.

    The MCP server record has the following tabs:

    1. Details: Displays all the asset details along with the MCP server URL at the bottom of the page.
    2. Approvals: Displays all the approval requests for the MCP server.
      Note:

      When MCP servers are synced to the AI Gateway with the AI Control Tower Core (with Now Assist), an option appears in AI Control Tower to start the approval process. However, with the AI Control Tower Pro license version, the button doesn’t appear after the synchronization job runs, and the Lifecycle status is shown as None.

      This behavior is known. As a workaround, verify that the lifecycle phase is set to New and the lifecycle status is set to AI steward review for the relevant sn_ai_governance_asset_governance_details record.

      Manually run a script to verify.

      var assetGovernanceRecord = new GlideRecord('sn_ai_governance_asset_governance_details');

if (assetGovernanceRecord.get('<sys_id of asset governance record>')) {

    assetGovernanceRecord.setValue('status', 5);

    assetGovernanceRecord.setDisplayValue('lifecycle_phase','New');

    assetGovernanceRecord.update();

}

      For more information on workflow of MCP server approval request, see Explore the MCP server approval workflow.

    3. KPIs & Metrics: Displays the MCP tools observability metrics for each server, which is approved and running. In the KPIs & Metrics tab, the AI steward can view different tools, which have been invoked across gateway usage from different hosts. The tab shows the total number of requests sent, the success rate, and any latency observed.
      Note:
      The AI Gateway tab in AI Control Tower tab on the AI Control Tower landing page also displays server usage. It shows the total number of requests or transactions and the corresponding success rate.
    4. AI Gateway setup: The AI Gateway setup tab appears only after an MCP server request is approved. From the AI Gateway setup tab, you can pause the AI Gateway transaction of the MCP server and add an MCP Client integration. The AI Gateway setup tab is accessible for all AI Control Tower workspace users.

      The AI Gateway setup tab has these sections:

      • AI Gateway MCP server details: Displays the following URLs:
        • AI Gateway MCP server URL: The Gateway endpoint that clients connect to.
        • Authorization endpoint URL: OAuth authorization endpoint.
        • Token endpoint URL: OAuth token exchange endpoint.
        Note:
        All these URLs are generated automatically and can’t be modified.

        If an MCP server is identified as malicious for a certain period, the AI stewards can pause all transactions directed to that server through the AI Gateway. Once the risk is resolved, transactions can be resumed. Similarly, at the AI Gateway level, it’s possible to pause all transactions across all servers from the AI Gateway page in the Configurations section of AI Control Tower.

      • MCP Client Integration: Displays Registered clients for the server. Each agent builder requires its own client registration and MCP client Integration generates a client ID and client secret. If you want to enable AI Agent Studio for a linear MCP server, you can create an MCP client integration by selecting Add.
        Note:
        For more information Gateway setup and MCP Client Integration, see Set up the AI Gateway and Client registration.
      • Global MCP Clients: Global MCP clients are set up separately from server records. You can configure them on the Configurations page of AI Control Tower in the AI Gateway section. Once they're created, they're automatically available across all MCP servers, so there's no need to set up individual client integrations for each server.
    5. Related Variants: The Related Variants tab provides a one-to-one mapping of all the MCP server entries from AI Agent Studio, so that each entry is visible and trackable in AI Control Tower. In the Related Variants tab, you can have multiple MCP servers of the same variant.
    Note:
    The AI steward can activate sensitive data protection for any transaction across servers by selecting Activate PII check. By activating, the AI gateway blocks those transactions, which involve sharing sensitive information to the MCP server.