MCP Server — LearnAgent on Azure Functions

MCP (Model Context Protocol) server deployed on Azure Functions that exposes a LearnAgent tool: any MCP-compatible client (Claude Code, VS Code, custom agents) can call it to search and retrieve official Microsoft Learn documentation in real time.

Built with Azure AI Foundry as the LLM backend. Two components:

1. agent_client.py — local agent that queries Microsoft Learn docs via MCP
2. server/ — MCP server deployed on Azure Functions, exposes a LearnAgent tool

---

Consuming the deployed MCP server

Endpoint:

https://azure-function-jnb-agents.azurewebsites.net/runtime/webhooks/mcp

Transport: Streamable HTTP (MCP protocol 2025-03-26)

Auth: Azure Functions system key in header:

x-functions-key: <mcp_extension_key>

Get the key:

az functionapp keys list \
  --name azure-function-jnb-agents \
  --resource-group NovilloBenitoJaime \
  --query systemKeys.mcp_extension -o tsv

Available tool

Tool	Input	Description
LearnAgent	query (string, required)	Searches and retrieves official Microsoft Learn documentation

Example — Python client

from mcp import ClientSession
from mcp.client.streamable_http import streamablehttp_client
import asyncio

KEY = "<mcp_extension_key>"
URL = "https://azure-function-jnb-agents.azurewebsites.net/runtime/webhooks/mcp"

async def main():
    async with streamablehttp_client(URL, headers={"x-functions-key": KEY}) as (r, w, _):
        async with ClientSession(r, w) as session:
            await session.initialize()
            result = await session.call_tool("LearnAgent", {"query": "How to deploy Azure Container Apps?"})
            print(result.content[0].text)

asyncio.run(main())

Install: pip install mcp

Example — VS Code / Claude Code (mcp.json)

{
  "mcpServers": {
    "LearnAgent": {
      "url": "https://azure-function-jnb-agents.azurewebsites.net/runtime/webhooks/mcp",
      "headers": {
        "x-functions-key": "<mcp_extension_key>"
      }
    }
  }
}

Example — curl (raw JSON-RPC)

KEY="<mcp_extension_key>"

# Initialize
curl -X POST https://azure-function-jnb-agents.azurewebsites.net/runtime/webhooks/mcp \
  -H "Content-Type: application/json" \
  -H "x-functions-key: $KEY" \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-03-26","capabilities":{},"clientInfo":{"name":"test","version":"1.0"}}}'

# Call tool
curl -X POST https://azure-function-jnb-agents.azurewebsites.net/runtime/webhooks/mcp \
  -H "Content-Type: application/json" \
  -H "x-functions-key: $KEY" \
  -d '{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"LearnAgent","arguments":{"query":"Azure Blob Storage Python quickstart"}}}'

---

Local agent (agent_client.py)

Connects to https://learn.microsoft.com/api/mcp directly and uses Azure AI Foundry as LLM.

Requirements: Python 3.9+, az login

python -m venv venv
venv\Scripts\activate
pip install -r requirements.txt

Create .env:

FOUNDRY_PROJECT_ENDPOINT=https://<resource>.services.ai.azure.com/api/projects/<project>
FOUNDRY_MODEL=gpt-4o-mini

python agent_client.py

---

Deploying the server

cd server
# Edit requirements.txt if needed
Compress-Archive -Path function_app.py,host.json,requirements.txt -DestinationPath deploy.zip -Force
az functionapp deployment source config-zip \
  --name azure-function-jnb-agents \
  --resource-group NovilloBenitoJaime \
  --src deploy.zip --build-remote true

Required app settings on the Function App:

Setting	Value
FOUNDRY_PROJECT_ENDPOINT	Azure AI Foundry project endpoint
FOUNDRY_MODEL	Model deployment name

The Function App needs System Assigned Managed Identity with Azure AI Developer role on the Foundry resource.

---

Stack

Component	Package
MCP transport	mcp>=1.9.0
Agent orchestration	agent-framework>=1.8.1
Azure Functions hosting	agent-framework-azurefunctions==1.0.0b260521
Azure AI Foundry LLM	agent-framework-foundry
Azure auth	azure-identity