{"id":11654,"date":"2026-01-07T08:00:39","date_gmt":"2026-01-07T16:00:39","guid":{"rendered":"https:\/\/devblogs.microsoft.com\/cosmosdb\/?p=11654"},"modified":"2026-01-07T09:12:34","modified_gmt":"2026-01-07T17:12:34","slug":"build-ai-tooling-in-go-with-the-mcp-sdk-connecting-ai-apps-to-databases","status":"publish","type":"post","link":"https:\/\/devblogs.microsoft.com\/cosmosdb\/build-ai-tooling-in-go-with-the-mcp-sdk-connecting-ai-apps-to-databases\/","title":{"rendered":"Build AI Tooling in Go with the MCP SDK \u2013 Connecting AI Apps to Databases"},"content":{"rendered":"
\n
A hands\u2011on walkthrough of building MCP servers that can plug AI applications into Azure Cosmos DB<\/p>\n

<\/strong><\/span><\/em><\/div>\n<\/div>\n

The\u00a0Model Context Protocol<\/a>\u00a0(MCP) has established itself as the ubiquitous standard for connecting AI applications to external systems. Since its release, there have been\u00a0implementations<\/a>\u00a0across various programming languages and frameworks, enabling developers to build solutions that expose data sources, tools, and workflows to AI applications.<\/p>\n

For Go developers, however, the journey to an official MCP SDK took longer (compared to other SDKs like Python and TypeScript). Discussions<\/a>\u00a0and\u00a0design\/implementation<\/a>\u00a0work on the official Go implementation began during early to mid 2025. At the time of writing (January 2026) it stands at version\u00a01.2.0<\/strong>. As a Gopher, I’m excited (and relieved!) to finally have a stable,\u00a0official MCP Go SDK<\/a>\u00a0that the Go community can rely on.<\/p>\n

To explore its capabilities, I built an\u00a0MCP server for Azure Cosmos DB<\/a>. This blog post will dive into the MCP Go SDK fundamentals by walking through it specifics, and exploring concepts such as tools, servers, etc. By the end, you’ll understand how to use the MCP Go SDK to build your own MCP servers, with Azure Cosmos DB serving as a practical example.<\/p>\n

Note:<\/strong>\u00a0This project is not intended to replace the\u00a0Azure MCP Server<\/a>\u00a0or\u00a0Azure Cosmos DB MCP Toolkit<\/a>. Rather, it serves as an experimental\u00a0learning tool<\/strong> that demonstrates how to combine the Azure and MCP Go SDKs to build AI tooling for Azure Cosmos DB.<\/p>\n

This project is not intended to replace the\u00a0Azure MCP Server<\/a>\u00a0or\u00a0Azure Cosmos DB MCP Toolkit<\/a>. Rather, it serves as an experimental\u00a0learning tool<\/strong>\u00a0that demonstrates how to combine the Azure Go SDK and MCP Go SDK Cosmos DB to build AI tooling for Azure Cosmos DB.<\/div><\/p>\n

MCP Basics<\/h2>\n

Let’s briefly cover what MCP is and how the MCP Go SDK works.<\/p>\n

What is the Model Context Protocol?<\/h3>\n

The Model Context Protocol (MCP) is an open-source standard for connecting AI applications to external systems. It’s often referred to as a USB-C port for AI applications<\/strong>\u00a0\u2014 just as USB-C provides a standardized way to connect devices, MCP provides a standardized way to connect AI applications to data sources, tools, and workflows.<\/p>\n

With MCP, AI applications (ranging from IDEs like VS Code, CLI coding tools like GitHub Copilot or apps like Claude web\/desktop) can:<\/p>\n

    \n
  • Access\u00a0data sources<\/strong>\u00a0(local files, databases, APIs)<\/li>\n
  • Use\u00a0tools<\/strong>\u00a0(search engines, calculators, external services)<\/li>\n
  • Execute\u00a0workflows<\/strong>\u00a0(specialized prompts, multi-step operations)<\/li>\n<\/ul>\n

    This standardization means developers can build MCP servers once and have them work with any MCP-compatible AI application, rather than creating custom integrations for each platform.<\/p>\n

    MCP Go SDK<\/h3>\n

    The\u00a0official Go MCP SDK<\/a>\u00a0provides the building blocks to create MCP servers and clients in Go. Here’s a minimal example of an MCP server with a simple tool:<\/p>\n

    package main\r\n\r\nimport (\r\n    \"context\"\r\n    \"log\"\r\n    \"strings\"\r\n\r\n    \"github.com\/modelcontextprotocol\/go-sdk\/mcp\"\r\n)\r\n\r\ntype ReverseInput struct {\r\n    Text string `json:\"text\" jsonschema:\"the text to reverse\"`\r\n}\r\n\r\ntype ReverseOutput struct {\r\n    Reversed string `json:\"reversed\" jsonschema:\"the reversed text\"`\r\n}\r\n\r\nfunc ReverseText(ctx context.Context, req *mcp.CallToolRequest, input ReverseInput) (\r\n    *mcp.CallToolResult,\r\n    ReverseOutput,\r\n    error,\r\n) {\r\n    runes := []rune(input.Text)\r\n    for i, j := 0, len(runes)-1; i < j; i, j = i+1, j-1 {\r\n        runes[i], runes[j] = runes[j], runes[i]\r\n    }\r\n    return nil, ReverseOutput{Reversed: string(runes)}, nil\r\n}\r\n\r\nfunc main() {\r\n    \/\/ Create server\r\n    server := mcp.NewServer(&mcp.Implementation{\r\n        Name:    \"text-tools\",\r\n        Version: \"v1.0.0\",\r\n    }, nil)\r\n\r\n    \/\/ Add a tool\r\n    mcp.AddTool(server, &mcp.Tool{\r\n        Name:        \"reverse\",\r\n        Description: \"reverses the input text\",\r\n    }, ReverseText)\r\n\r\n    \/\/ Run over stdio\r\n    if err := server.Run(context.Background(), &mcp.StdioTransport{}); err != nil {\r\n        log.Fatal(err)\r\n    }\r\n}<\/code><\/pre>\n
    This example demonstrates the key concepts:<\/p>\n
      \n
    • Tool definition<\/strong>: A\u00a0mcp.Tool<\/code>\u00a0with a name and description<\/li>\n
    • Input\/Output types<\/strong>: Structs with JSON schema tags that define the tool’s interface<\/li>\n
    • Handler function<\/strong>: The actual logic that executes when the tool is called<\/li>\n
    • Server<\/strong>: Created with\u00a0mcp.NewServer()<\/code>\u00a0and configured with tools<\/li>\n
    • Transport<\/strong>: How the server communicates (here using stdio)<\/li>\n<\/ul>\n
      These concepts will be covered later on the blog.<\/p>\n
      MCP Server in Action<\/h2>\n
      \u25b6\ufe0f To get a sense of what the MCP server can do, take a look at this short demo of using the server with Agent Mode in Visual Studio Code<\/a>:<\/p>\n