This practical guide explains what it is, why it matters, and how to integrate Azure AI Agents Service with Claude Desktop using the Model Context Protocol (MCP). Although this guide specifically focuses on Claude Desktop, the concepts can be extended to any MCP-compatible client.

What is Model Context Protocol (MCP)? Copy link

Model Context Protocol (MCP) is an open standard designed to seamlessly connect AI assistants with diverse data sources, enabling context-aware interactions. MCP allows AI applications to intelligently retrieve, process, and leverage information across private document repositories and public web sources, significantly enhancing their functionality.

Azure AI Agent Service is a fully managed service designed to empower developers to securely build, deploy, and scale high-quality, and extensible AI agents without needing to manage the underlying compute and storage resources.

Model Context Protocol (MCP) is now integrated with Azure AI Agent Service. MCP creates a common language that lets AI models dynamically use agents that have access to knowledge and tools. MCP combined with Azure AI Agent Service, gives knowledge retrieval from public and private data sources, including:

• Real-time web data using Grounding with Bing Search
• Internal, private data with Azure AI Search
• and more sources like SharePoint and Fabric coming soon…

Prerequisites Copy link

Ensure you have:

• Python 3.10+
• Claude Desktop or another MCP-compatible client
• Azure CLI (az) configured
• Existing Azure AI Agents configured in Azure AI Foundry

Step-by-Step Integration Copy link

Step 1: Azure Configuration

• Configure your Azure AI Agents in Azure AI Foundry.
• Retrieve your Azure AI Project connection string and agent IDs.
• Authenticate Azure CLI:
  Copy

  az login

Step 2: Setting Environment Variables

Create an .env file in your project’s root:

PROJECT_CONNECTION_STRING=your-project-connection-string
DEFAULT_AGENT_ID=your-default-agent-id  # optional

Step 3: Server Installation and Execution

Set up a virtual environment and install dependencies:

uv venv
source .venv/bin/activate  # macOS/Linux
.venv\Scripts\activate     # Windows
uv add mcp[cli] azure-identity python-dotenv azure-ai-projects aiohttp

Start the MCP server:

uv run -m python.azure_agent_mcp_server

Step 4: Configure Claude Desktop (or another MCP client)

Update your MCP client configuration (claude_desktop_config.json) to integrate the MCP server:

{
  "mcpServers": {
    "azure-agent": {
      "command": "uv",
      "args": [
        "--directory",
        "/ABSOLUTE/PATH/TO/PARENT/FOLDER",
        "run",
        "-m",
        "azure_agent_mcp_server"
      ],
      "env": {
        "PROJECT_CONNECTION_STRING": "your-project-connection-string",
        "DEFAULT_AGENT_ID": "your-default-agent-id"
      }
    }
  }
}

Practical Usage Copy link

Your MCP server provides several useful methods and can leverage various tools configured with your agents, including Azure AI Search and Bing Web Grounding Tools:

• Connect to a specific Azure AI agent:
  Copy

  connect_agent(agent_id="agent-1234", query="What's the forecast today?")

• Query the default configured agent:
  Copy

  query_default_agent(query="Summarize recent updates.")

• List available agents:
  Copy

  list_agents()

Each MCP client interaction maintains isolated threads for privacy and coherent conversations.

Real-World Impact Copy link

Integrating Azure AI Agent Service with MCP-compatible clients like Claude Desktop:

• Enhances productivity by directly embedding conversational AI capabilities into desktop workflows.
• Provides scalable, secure AI interactions without extensive custom integration.
• Reduces development complexity with standardized MCP-based connections.

Conclusion Copy link

By using MCP, developers can significantly extend the capabilities of desktop clients, integrating powerful Azure AI Agents easily and effectively. Though this guide highlights Claude Desktop, the same MCP integration principles apply broadly, making it a powerful tool for enhancing AI-driven application workflows. For .NET developers, explore how you can leverage MCP within your projects using Semantic Kernel. Check out this step-by-step guide to learn more.

We invite you to explore further customizations and share your feedback or questions!