Set up Microsoft Copilot Studio with Remote MCP

Scenario

Integrating the OvalEdge Remote MCP (Model Context Protocol) Server with Microsoft Copilot Studio enables Microsoft 365 Copilot (chat and desktop applications) and Microsoft Teams users to securely access governed metadata from OvalEdge through AI-assisted conversations. After the integration is complete, users can search catalog assets, retrieve metadata, explore glossary terms, analyze lineage, discover data sources, and access other supported MCP capabilities directly from Microsoft 365 Copilot.

Prerequisites

Before starting the integration, ensure that:

• Microsoft Copilot Studio access is available.

• The OvalEdge Remote MCP Server URL is available.

• The OvalEdge Token and Secret have been generated.

• The authentication value is available in the format TOKEN::SECRET.

• Permission is available to create and publish Microsoft Copilot agents.

Steps Involved

1. Open Microsoft Copilot Studio and sign in with a Microsoft account that has permission to create Copilot agents.

   
2. From the left navigation pane, click Agents. Click Create blank agent.

   
3. Enter the Agent Name, and then click Create.

   
4. Wait until Microsoft Copilot Studio completes provisioning the new agent. Select the preferred AI model for the agent.

   

   By default, the Claude Sonnet 4.6 AI model is selected.

5. Click the Tools tab. Click Add Tool.

   
6. Select Model Context Protocol (MCP) as the tool type.

   
7. Enter the Server Name, Server Description, and OvalEdge Remote MCP Server URL for the MCP server.

   
8. Select API Key as the authentication type. Select Header as the API Key location. Enter X-OvalEdge-Credentials as the header name. Click Create to register the MCP server.

   
9. Verify that the newly created MCP server appears as below.

   
10. Click Create New Connection for the MCP server.

    
11. Enter the authentication value using the format TOKEN::SECRET.

    
12. Click Create to establish the MCP connection.

13. Verify that the connection status changes to Connected after the connection is created.

    
14. Click Add and Configure to retrieve the tools exposed by the OvalEdge MCP Server.

15. Review the available MCP tools displayed in the tool list.

    

    By default, all MCP tools are selected, but users can customize it to enable only the required tools and disable those that are not required. Only selected MCP tools will be available to the agent.

16. Click Save to save the selected MCP tools.

    
17. Navigate to Settings for the agent.

    
18. Select Generative AI. Click Orchestration. Enable Use Generative AI Orchestration by setting it to Yes.

    
19. Navigate to the Test panel. Enter a sample prompt to verify the MCP integration. Click Send to submit the prompt.

    
20. If prompted to authenticate, click Open Connection Manager.

    
21. Review the available MCP connections. Verify that the OvalEdge MCP connection is listed.

22. If the connection is not active, select the OvalEdge MCP connection. Click Connect.

    
23. Select the required connection and click Submit.

    
24. Wait until the connection status changes to Connected.

    
25. Return to the Test panel. Click Retry to execute the original prompt again.

    
26. Verify that the Activity Map displays the MCP tool execution. Review the tool execution details, including the request and response payload.

    

    Verify that the execution status is successful and no errors are returned. Review the metadata returned from the OvalEdge MCP Server. Confirm that the returned metadata matches the requested catalog object.

27. Navigate to the Channels tab. Under Microsoft Channels, select Microsoft 365 and Microsoft Teams.

    
28. Enable Make agent available in Microsoft 365 Copilot. Click Add Channel.

    
29. Verify that the ‘Microsoft 365 and Microsoft Teams’ channel is successfully added to the agent.

    
30. Click Publish. In the Publish Agent dialog, enable Force newest version.

    
31. Click Publish to publish the latest version of the agent.

32. Wait until the publishing process completes successfully. Verify that the agent status changes to Published.

    
33. Navigate to Microsoft 365 and Microsoft Teams under the Channels page.

    
34. Click See agent in Microsoft 365 to open the published agent directly in Microsoft 365 Copilot.

    
35. Verify that the published OvalEdge MCP agent opens successfully in Microsoft 365 Copilot. In Microsoft 365 Copilot, review the Agent Store and locate the published OvalEdge MCP agent.

    
36. Click Add to add the OvalEdge MCP agent to Microsoft 365 Copilot.

    
37. Wait until the agent is added successfully.

38. Open the newly added OvalEdge MCP agent.

    

    Users can pin the created Agent(s), and the pinned Agent(s) will be displayed under the ‘Agents’ section from the left menu.

39. Enter a sample metadata query in the prompt box. Click Send to submit the query.

    

    Ensure that users select only the created agent from the ‘Agents’ section (here ovaledgeMCP2 Agent) before prompting.

40. If prompted for authentication, click Open Connection Manager.

    
41. In the Manage your Connections page, locate the OvalEdge MCP connection.

    
42. Click Connect for the OvalEdge MCP connection.

    
43. In the Create or Pick a Connection dialog, select the previously created OvalEdge MCP connection.

    
44. Click Submit to authenticate the connection.

45. Wait until the connection status changes to Connected.

    
46. Return to the Microsoft 365 Copilot conversation window. Click Retry to execute the previously submitted prompt.

    
47. Wait for Microsoft 365 Copilot to invoke the OvalEdge MCP tools.

48. Review the returned response and verify that the metadata has been successfully retrieved from the OvalEdge MCP Server.

    

Copyright © 2026, OvalEdge LLC, Peachtree Corners, GA, USA.