HOPEX MCP Server (Beta)

Summary

This MCP (Model Context Protocol) connector exposes the HOPEX repository through an AI-native interface that can be queried by AI agents and assistants. The connector follows the MCP specification framework, allowing AI models to interact naturally with HOPEX data including applications, business processes, capabilities, and their relationships. With this connector, AI agents can read and analyze enterprise architecture data from the HOPEX repository in real-time.

Key Features

• AI-Native Access: Direct integration with AI agents through the Model Context Protocol (MCP), enabling natural language queries of HOPEX repository data
• GraphQL-Based Architecture: Leverages GraphQL specification for flexible, efficient data retrieval with support for complex queries and relationships
• Comprehensive Schema Coverage: Access to main standard HOPEX types including Applications, Business Processes, Capabilities, Projects, Software Technologies, and their relationships
• Advanced Query Capabilities: Support for introspection, filtering, aggregation, counting, and multi-hop relationship traversal
• Path Finding: Intelligent relationship discovery between different types of objects in the repository
• Secure API Access: API key authentication protecting your HOPEX instance data
• Fine-Tunable AI Integration: Compatible with AI model parameters and configurations optimized for Copilot scenarios

What Hopex Does:

• Expose your data via MCP (Model Context Protocol), making it accessible to AI applications
• Handle secure API token-based authentication
• Maintain the data connection layer
• Provide a standard GRAPHQL schema based on our standard Meta model

What Hopex Does not:

• Build or provide AI agents such as Claude or ChatGPT (MCP does not connect to any LLM unless client decides to connect its own)
• Configure how clients should use the data
• Advise on optimal agent setup or prompting strategies as they depend on their data, etc.
• Create a customized graphql schema for client customizations

Requirements

• Active HOPEX instance with appropriate functional licenses
• AI agent or assistant that supports MCP (Model Context Protocol) connections
• AI agent must support API key authentication
• Network access to your HOPEX instance URL

Connection Configuration

Endpoint URL Format: https://<your-instance-name>/mcp

Replace <your-instance-name> with your actual HOPEX instance domain.

Authentication: API key-based authentication required. Configure your AI agent with:

• Connection type: MCP with API key
• API key: Provided by your HOPEX administrator

AI Model Parameters: For optimal performance with Copilot and similar AI assistants, configure model parameters according to recommended settings (temperature, token limits, etc.)

Example of connection declaration for Claude Desktop:

"<name-of-your-server-it-can-be-what-you-want>" : {
    "command": "npx",
    "args": [
        "mcp-remote",
        "<your-instance-name>/mcp",
        "--header",
        "x-api-key:<your-api-key>",
        "--header",
        "x-schema:<your-specific-schema>"
    ]
}

Depending on the adressed AI client here is another connection declaration (this one does not work for all AI clients):

"<name-of-your-server-it-can-be-what-you-want>": {
    "type": "http",
    "url": "<your-instance-name>/mcp",
    "headers": {
        "x-api-key": "<your-api-key>",
        "x-schema": "<your-specific-schema>"
    }
}

The "x-schema" parameter is optional as a default schema is defined in the module settings. The "x-api-key" parameter can be removed if a default api key is defined in the module settings.

For a local usage only:

If you use the MCP server and the AI client locally on your computer, the connection may not work. This is due to the fact that AI clients use nodeJs to interact with the MCP server and nodeJs use its own certificate store. To make it work you will need to define an environment variable called "NODE_OPTIONS" and set its value to "--use-system-ca" (it indicates to nodeJs to use the windows certification store instead of its own store).

Dependencies

• HOPEX Platform (compatible version with MCP support)
• AI agent supporting MCP protocol with API key authentication

Documentation

• MCP Protocol Specification: https://modelcontextprotocol.io/
• HOPEX Platform Documentation: Available through your HOPEX instance
• GraphQL Query Examples: Available within the connector introspection too

Release Notes

New

• Initial release of HOPEX MCP connector
• Support for most core HOPEX types (Application, Business Capability, Process, Project, Software Technology, etc.)
• GraphQL introspection capabilities for schema discovery
• Advanced query execution with filtering, ordering, and pagination
• Object counting with relationship-based filtering
• Path finder for discovering relationships between types
• API key-based security implementation
• Compatible with major AI platforms supporting MCP (Claude, ChatGPT, etc.)

Changed

• NA

Fixed

• NA

Removed

• NA