AI Backend

MCP Server Development

Model Context Protocol servers for integrating data sources and tools with AI agents.

Duration Laufend
MCP Server architecture

About the Project

The Model Context Protocol (MCP) enables AI agents standardized access to external tools and data sources. As part of various projects, several specialized MCP servers were developed that provide reusable integrations for typical enterprise data sources.

Custom Web Search

Custom Web Search MCP Server: An MCP server that encapsulates the optimized web search logic from the AI Research project. The component collects search results, embeds them in-memory and performs semantic searches – token-optimized chunks are returned to the requesting agent.

ERP/CRM Tools

ERP/CRM Tools MCP Server: An MCP server for bidirectional integration with enterprise systems – reading and writing in SQL-based ERP systems, reading and writing via REST APIs in CRM systems, structured tool definitions for typical CRUD operations.

SharePoint

SharePoint MCP Server: Comprehensive tool collection for SharePoint interaction – access to sites, lists and document libraries, document search and retrieval, metadata operations, integration via Microsoft Graph API.

Outlook

Outlook MCP Server: Tools for email integration – reading emails and attachments, creating and sending emails, calendar operations.

Highlights

  • Reusable MCP servers for typical enterprise integrations
  • Standardized protocol for agent-tool communication
  • Secure authentication via Managed Identity
  • Modularly extensible for additional data sources

Challenges

Balance between flexibility and security. Consistent API abstraction across different data sources.

Results

Reusable server collection for rapid agent integrations. Standardized enterprise tool connection.

Interested?

Are you interested in a similar project or want to learn more? Get in touch – we look forward to hearing from you!

Get in touch