Sure, you can handle it. But should you?
Let our experts manage the tech maintenance while you focus on your business.
Let our experts manage the tech maintenance while you focus on your business.
Whitepaper
Using the Mindbreeze InSpire MCP Server
Activating and configuring the MCP Server
Overview
The MCP Server exposes the built-in API tools of the Insight Services for RAG for the Search, Suggest, Personalization and Preview API respectively, so they can be used by an AI application.
Setting it up involves three stages:
- Enabling the server
- Publishing the tools you want to expose
- Connecting your AI application
Prerequisites
To setup the Mindbreeze InSpire MCP Server, access to the Insight Services for RAG service must be available. If this service is not yet configured, go to the Mindbreeze Management Center, open the menu “Configuration”, go to the tab “Indices” and configure the service in the section “Services”. For more information, see Configuration - InSpire AI Chat and Insight Services for Retrieval Augmented Generation.
After the Insight Services for RAG service is created and Insight Services for RAG is visible in the Mindbreeze InSpire Management Center, the steps described in the next chapter can be performed.
Step 1: Enable the MCP Server
In the Management Center, open the menu “Configuration”, go to the tab “Indices”, activate the “Advanced Settings” and expand the Insight Services for RAG service (“RAG Administration” in the screenshot below).
Go to the section “Agentic Pipelines” and locate the advanced setting “Enable MCP Server”. To enable the MCP server, make sure that this setting is enabled.
For more information, see Configuration - InSpire AI Chat and Insight Services for Retrieval Augmented Generation - Agentic Pipelines, MCP Servers and Tools.
Once enabled, the MCP Server listens at:
https://<your-inspire-host>:<client-service-port>/api/modelcontextprocol.2025-11-25/mcp
Example – if your client service listens on port 443 (default):
https://<your-inspire-host>/api/modelcontextprotocol.2025-11-25/mcp
Step 2: Configuring API Tools
Once the MCP Server is enabled, a new “Context” menu appears in Insight Services for RAG. Use it to manage MCP Tools and server behavior.
For more information, see Whitepaper - Administration of Insight Services for Retrieval Augmented Generation - Context.
Step 3: Enable Tools on the MCP Server
In the “MCP Server” section:
- Enable/disable the MCP Server itself – when disabled, no incoming requests are processed, regardless of tool state.
- Enable/disable individual published tools – controls which tools are actually exposed to AI applications.
For more information, see Whitepaper - Administration of Insight Services for Retrieval Augmented Generation - MCP Server.
Summary Checklist
To fully activate and use the MCP Server:
- Enable MCP Tools in the Insight Services for RAG configuration
- Publish the tools you want to expose
- Enable those published tools in the MCP Server configuration
- Connect your AI application to the MCP Server
Test and Development using the Mindbreeze InSpire MCP Server
For testing or developing with Mindbreeze InSpire MCP Server one can use an arbitrary development tool, like the popular MCP inspector (https://modelcontextprotocol.io/docs/tools/inspector).
In the following we show a development workflow example using authenticated requests with the MCP inspector tool.
Authentication
If the Mindbreeze InSpire MCP server endpoint requires OAuth JWT Token authentication, it is required that an access token is sent with the MCP requests. An access token can be manually required from the configured OAuth authorization server’s token endpoint.
Here is an example curl request where an access token is required from a Microsoft Entra ID application using the username-password authentication flow:
curl -X POST https://login.microsoftonline.com/{{tenant_id}}/oauth2/v2.0/token \ -H "Content-Type: application/x-www-form-urlencoded" \ -d 'client_id={{client_id}}' \ -d 'scope={{application_scope_uri}} ' \ -d 'username={{username}}' \ -d 'password={{password}} \ -d 'grant_type=password' \ -d 'client_secret={{client_secret}}' |
The placeholders used in the example should be set to the following values:
{{tenant_id}} | Directory (tenant) ID of the Microsoft Entra ID application |
{{client_id}} | Application (client) ID of the Microsoft Entra ID application |
{{client_secret}} | Client secret of the Microsoft Entra ID application |
{{application_scope_uri}} | Full URI of the defined scope for the Microsoft Entra ID application |
{{username}} | Username of the Microsoft Entra ID user on behalf of the MCP client is acting |
{{password}} | Password of the Microsoft Entra ID user on behalf of the MCP client is acting |
Exploring the MCP Server with MCP Inspector
Setting the Authentication Header
If the Mindbreeze InSpire MCP Server requires authentication via JWT, and you acquired an ID token, this token can be used by setting the “Authorization” header in MCP Inspector to the following value:
Bearer <jwt_token> |
<jwt_token> being the Base64 encoded JWT ID token as you receive from the authorization server’s token endpoint.
Connect to the InSpire MCP endpoint
- Set Transport Type to ‘Streamable HTTP’
- Set URL to (ensure the MCP-Server is enabled)
- Click Connect.
Verify the connection
A successful connection shows the server as connected, and the Tools, Resources, and Prompts tabs are populated with the capabilities exposed by the InSpire MCP endpoint. Use List Tools to confirm the endpoint responds.
