5. MCP Integration

5. MCP Integration

In this module, you connect OpsAgent to a hosted Model Context Protocol (MCP) server from Microsoft Learn so the agent can answer questions with up-to-date documentation context.

Module Goals

By the end of this module, you will be able to:

Understand how hosted MCP tools work with Microsoft Agent Framework
Create an OpsAgent that uses only an MCP tool (no local function tools)
Run one user query and return one grounded answer from MCP-backed retrieval

In This Module

What Is MCP Integration
Step 1 through Step 5
Expected Outcomes

What Is MCP Integration

MCP integration lets an agent call tools exposed by a remote MCP server endpoint. In this module, the MCP server is hosted by Microsoft Learn at https://learn.microsoft.com/api/mcp.

With MCP Integration:

The agent can query external knowledge through the MCP tool
You can connect generic tools from standard MCP servers securely
The framework manages execution and parsing seamlessly

MCP Flow

flowchart TD
    U([User Query]) --> A[OpsAgent]
    A --> M[Microsoft Learn MCP Tool]
    M --> D[Microsoft Learn Content]
    D --> A
    A --> O([Grounded Answer])

Step 1 - Create the Test Script

Inside lab/test_mcp_integration.py, add the script for this module.

touch test_mcp_integration.py

Step 2 - Ensure GitHub Environment Variables

Make sure you have your GitHub connection values in .env (carried over from the previous modules):

GITHUB_TOKEN=github_pat_...
GITHUB_MODEL=gpt-4o-mini

Step 3 - Create the MCP Tool (MCP Only)

Create the MCPStreamableHTTPTool for Microsoft Learn content.

from agent_framework import MCPStreamableHTTPTool

# MCP-only setup: no local Python function tools are added.
learn_mcp = MCPStreamableHTTPTool(
    name="Microsoft Learn MCP",
    url="https://learn.microsoft.com/api/mcp",
    approval_mode="never_require",
)

Step 4 - Register MCP Tool and Run One Query

As before, register the MCP tool in Agent(tools=[...]). The setup of the AsyncOpenAI client uses the GitHub Models endpoint just like in Module 4.

async with Agent(
    client=client,
    name="OpsAgent",
    description="OpsAgent is an AI-powered operations and engineering assistant.",
    instructions=(
        "You are OpsAgent, an AI-powered operations and engineering assistant. "
        "Answer using Microsoft Learn MCP results only, with concise, actionable guidance for cloud operations."
    ),
    tools=[learn_mcp],
) as agent:
    result = await agent.run(query)
    print(f"\n💬 OpsAgent: {result.text}\n")

Step 5 - Complete Code (Single File)

"""
Module 5 - Test MCP Integration with OpsAgent

Run:
    python test_mcp_integration.py
    or
    uv run test_mcp_integration.py
"""

import asyncio
import os

from dotenv import load_dotenv
from openai import AsyncOpenAI

from agent_framework import Agent, MCPStreamableHTTPTool
from agent_framework.openai import OpenAIChatCompletionClient

load_dotenv()

github_base_url = "https://models.github.ai/inference"
GITHUB_TOKEN = os.getenv("GITHUB_TOKEN")
GITHUB_MODEL = os.getenv("GITHUB_MODEL")

if not GITHUB_TOKEN or not GITHUB_MODEL:
    raise ValueError("GITHUB_TOKEN and GITHUB_MODEL must be set in the .env file")


async def main():
    """Create and run OpsAgent with a hosted Microsoft Learn MCP tool."""

    print("🤖 OpsAgent (MCP) is ready for one question.\n")

    try:
        query = input("👤 You: ").strip()
    except (EOFError, KeyboardInterrupt):
        print("\n👋 Goodbye!")
        return

    if not query:
        print("No query provided. Exiting.")
        return

    async_openai = AsyncOpenAI(
        api_key=GITHUB_TOKEN,
        base_url=github_base_url,
    )

    client = OpenAIChatCompletionClient(
        model=GITHUB_MODEL,
        async_client=async_openai,
    )

    # MCP-only setup: no local Python function tools are added.
    learn_mcp = MCPStreamableHTTPTool(
        name="Microsoft Learn MCP",
        url="https://learn.microsoft.com/api/mcp",
        approval_mode="never_require",
    )

    async with Agent(
        client=client,
        name="OpsAgent",
        description="OpsAgent is an AI-powered operations and engineering assistant.",
        instructions=(
            "You are OpsAgent, an AI-powered operations and engineering assistant. "
            "Answer using Microsoft Learn MCP results only, with concise, actionable guidance for cloud operations."
        ),
        tools=[learn_mcp],
    ) as agent:
        result = await agent.run(query)
        print(f"\n💬 OpsAgent: {result.text}\n")


if __name__ == "__main__":
    asyncio.run(main())

Run the Script

python test_mcp_integration.py
# or
uv run test_mcp_integration.py

Example Output

🤖 OpsAgent (MCP) is ready for one question.

👤 You: What is Azure App Service?

💬 OpsAgent: Azure App Service is a fully managed Platform as a Service (PaaS) that enables users to build, deploy, and scale web apps, mobile app back ends, and RESTful APIs without needing to manage the underlying infrastructure...

Expected Outcomes

OpsAgent initializes with GitHub Models credentials
The agent uses only one MCP tool: Microsoft Learn MCP
A single user query is processed and answered once
The response is grounded by MCP-backed documentation retrieval