--- sourceDocument: Australia Build or modify applications sourceDocumentLink: https://www.servicenow.com/docs/r/australia/application-development Release : - australia ft:locale : - en-US ft:publication_title : - Australia Build or modify applications ft:clusterId : - cadev bundleId : - cadev workflow : - Development, Data, and Analytics --- # MCP connections # MCP connections and Autonomous Engineer {#ariaid-title1} Release version: Australia Updated September 23, 2026 ![](https://www.servicenow.com/docs/portal-asset/ico-clock) 3 minutes to read Summarize ![AI sparkle icon](https://servicenow.com/docs/portal-asset/ai-sparkle-icon) 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 connections and Autonomous Engineer MCP connections enable Autonomous Engineer to integrate with external tools and systems using a standardized communication protocol called Model Context Protocol (MCP). This open protocol allows AI agents like Autonomous Engineer to discover, invoke, and share data with third-party applications without the need for custom integration logic. MCP connections extend Autonomous Engineer's planning and implementation workflows by incorporating external data and capabilities seamlessly. Show full answer Show less ## Key Features * **Unified Connection Registry:** Autonomous Engineer shares MCP connections with Build Agent, so connections approved and authenticated for one are available to the other without separate configuration. * **Workflow Integration:** MCP connections can be used throughout Autonomous Engineer's workflow---from reading project requirements during planning to creating or updating records in connected systems during implementation. * **Approval and Activation Process:** Each MCP server must be approved as an AI asset by a user with the `snaigovernance.aisteward` role in AI Control Tower before it can be enabled. After approval, users authenticate connections via Personal Integrations and enable the MCP server in Build Agent settings. * **Supported MCP Servers:** Autonomous Engineer currently supports MCP connections to popular third-party applications such as Atlassian, AWS DevOps, Box, Docusign, Figma, Linear, Miro, Prisma, Postgres, and Zoom (including Chat, Docs, and Whiteboard). Custom MCP servers can also be created on the ServiceNow AI Platform. * **Tool Visibility:** Available tools for each MCP server can be viewed in the MCP servers tab within the Build Agent settings panel, which is shared with Autonomous Engineer. ## How to Use and Configure MCP Connections * An administrator adds the MCP server as a Workflow Data Fabric connection. * A user with the appropriate governance role approves the MCP server in AI Control Tower. * Users authenticate the connection via the Personal Authentication dashboard. * The MCP server is enabled in Build Agent settings to make it available for use by Autonomous Engineer. * Configuration methods include dynamic client registration (one-click), manual client registration, or using a custom connector for servers not listed in the MCP registry. Because Autonomous Engineer uses the same settings panel as Build Agent, configuration steps are unified, simplifying management of MCP connections across products. MCP connections enable Autonomous Engineer to access external tools and resources through standardized communication, extending the planning and implementation workflow with data and context from third-party systems. MCP is an open protocol that defines how AI agents communicate with external systems to discover and invoke tools, retrieve resources, and share context. By standardizing this communication layer, MCP enables AI agents to interact with a wide range of third-party applications without requiring custom integration logic for each one. In Connect Hub, an MCP connector represents a configured connection between ServiceNow and an external system that exposes a server compatible with MCP. After an administrator sets up an MCP connector and you authenticate your connection, Autonomous Engineer can use it to access tools and capabilities from the external system. This enables Autonomous Engineer to pull in external context during planning and to create records in connected systems as part of implementation. ## How Autonomous Engineer uses MCP connections {#ae-mcp-connections__section_ae_mcp_relationship} Autonomous Engineer and Build Agent share the same MCP connection registry. Connections that are approved, authenticated, and enabled for Build Agent are available to Autonomous Engineer in the same session. You don't have to configure connections separately for each product. Autonomous Engineer can use MCP connections at any stage of the workflow. For example, during planning, Autonomous Engineer can read requirements from a connected project management tool. During implementation, it can create or update records in connected systems as part of executing a plan. ## Approve and activate MCP servers {#ae-mcp-connections__section_ae_mcp_prereqs} Before you can enable an MCP server in Build Agent, a user with the sn_ai_governance.ai_steward role must approve it as an AI asset in AI Control Tower. An administrator without this role can't approve the server. Each MCP server requires this approval, regardless of whether the server is enabled by default. For the approval steps, see [Onboard MCP server approval playbook](https://www.servicenow.com/docs/access?context=aict-onboard-mcp-server-approval-playbook&version=australia&pubname=australia-intelligent-experiences&ft:locale=en-US). The end-to-end flow for making an MCP server available is: 1. The administrator adds the MCP server as a Workflow Data Fabric (WDF) connection. 2. A user with the sn_ai_governance.ai_steward role approves the server as an AI asset in AI Control Tower. 3. You authenticate the connection in Personal Integrations. For details, see [Using the Personal Authentication dashboard](https://www.servicenow.com/docs/access?context=personal-auth-dashboard&version=australia&pubname=australia-integrate-applications&ft:locale=en-US). 4. You enable the MCP server in Build Agent settings. {#ae-mcp-connections__ol_scy_rw4_njc}For details on adding the MCP server connection, see the following topics: * [Dynamic client registration](https://www.servicenow.com/docs/access?context=mcp-dcr-a&version=australia&pubname=australia-integrate-applications&ft:locale=en-US) for servers that connect with one click * [Manual client registration](https://www.servicenow.com/docs/access?context=mcp-mcr-a&version=australia&pubname=australia-integrate-applications&ft:locale=en-US) for servers marked Manual setup may be required * [Client registration using custom connector](https://www.servicenow.com/docs/access?context=mcp-custom&version=australia&pubname=australia-integrate-applications&ft:locale=en-US) for servers that aren't in the MCP registry {#ae-mcp-connections__ul_ba_mcp_registration_1} Individual MCP servers are enabled by default, but the complete flow must be completed before any server is available for use. For details on adding a new MCP connection in Workflow Data Fabric, see [Model Context Protocol connectors](https://www.servicenow.com/docs/access?context=model-context-protocol-connector&version=australia&pubname=australia-integrate-applications&ft:locale=en-US). ## Supported MCP servers {#ae-mcp-connections__section_ae_mcp_supported} The following MCP connections are currently supported for Autonomous Engineer: * Atlassian Rovo * AWS DevOps * Box * Docusign * Figma * Linear * Miro * Prisma Postgres * Zoom * Zoom Chat * Zoom Docs * Zoom Whiteboard {#ae-mcp-connections__ul_ltk_db5_ykc} Note: Because Build Agent doesn't support multiple servers connecting to one app, each Zoom integration requires its own app in Zoom. You can create custom MCP servers on the ServiceNow AI Platform and then have Build Agent connect to them. For information about creating custom MCP servers, see [Create a Model Context Protocol server](https://www.servicenow.com/docs/access?context=create-mcp-server&version=australia&pubname=australia-intelligent-experiences&ft:locale=en-US). ## View available tools per MCP server {#ae-mcp-connections__section_ae_mcp_view_tools} You can view the tools available for each connected MCP server in the MCP servers tab in the Build Agent settings panel. Expand a server by selecting the chevron next to the server name to see its tools and descriptions. Because Autonomous Engineer runs within the Build Agent interface, the same pane applies when Autonomous Engineer is the active mode. ## How to configure MCP connections {#ae-mcp-connections__section_ae_mcp_config} For details on configuring MCP connections, see [Connect Build Agent to a supported MCP server](https://www.servicenow.com/docs/a7jVLFOgqGPLT0iMDDiSRQ "Connect a supported MCP server to Build Agent to access external tools and resources in the chat panel when building and editing apps."). Because Autonomous Engineer shares the Build Agent settings panel, the configuration steps are the same for both products. For more information on MCP connections on the platform, see [Additional connector configurations](https://www.servicenow.com/docs/access?context=additional-configs-mcr&version=australia&pubname=australia-integrate-applications&ft:locale=en-US).