Azure FastMCP Server

A FastMCP (Model Context Protocol) server that exposes Azure discovery, inventory, and cost insights as MCP tools and resources. This server allows LLMs and MCP-compatible clients to safely query Azure Resource Graph, Subscription metadata, and Azure Cost Management using Service Principal authentication.

The server is designed to be deployed:

• Locally (Python)
• As a Docker container
• In FastMCP Cloud

What This Server Does

This FastMCP server provides:

Azure Discovery & Inventory

• Query Azure Resource Graph across one or more subscriptions
• Fetch full inventories or drill into specific resource types
• Retrieve metadata for a single resource by ARM ID
• Describe a subscription with high-level resource rollups

Azure Cost Insights

• Query Azure Cost Management
• Support daily or monthly granularity
• Group costs by resource, resource group, or service
• Filter by resource ID
• Support predefined or custom time ranges

MCP Resources for LLM Context

• Azure Resource Graph KQL syntax documentation
• Azure Resource Graph supported tables & schemas

Utility Tools

• Schema search for finding correct ARG table names
• Server metadata and health endpoint

Exposed MCP Tools

fetch_azure_resources

Query Azure Resource Graph across one or more subscriptions.

Supports:

• Inventory queries
• Queries by resource ID
• Fully custom KQL queries

describe_subscription

Returns:

• Subscription metadata
• Total resource counts
• Resource breakdown by category and type

fetch_costs

Fetch Azure cost data with:

• Daily or monthly granularity
• Grouping options
• Resource-level filtering

search_azure_resource_schema

Search supported Azure Resource Graph tables and resource types.

get_kql_syntax_guide

Returns official Azure Resource Graph KQL documentation.

Utility Tools

• get_server_info

MCP Resources

azure://kql/syntax

Official Azure Resource Graph KQL syntax documentation.

azure://schema/tables

List of supported Resource Graph tables and resource types.

Authentication Model

This server only supports Azure Service Principal authentication.

The following environment variables are required:

AZURE_TENANT_ID=<tenant-id>
AZURE_CLIENT_ID=<app-id>
AZURE_CLIENT_SECRET=<client-secret>

The Service Principal must have:

• Reader access for Resource Graph queries
• Cost Management Reader for cost queries

Health Endpoint

The server exposes a health endpoint used by Docker and cloud platforms:

GET /health

Response:

{
  "status": "healthy",
  "service": "fastmcp-server",
  "version": "1.0.0"
}

Running Locally (Python)

Prerequisites

• Python 3.12+
• Azure Service Principal credentials

Install Dependencies

pip install -r requirements.txt

Run the Server

python server.py

The server will start on:

http://localhost:8000

Docker Deployment

Build the Image

docker build -t azure-fastmcp-server .

Run the Container

docker run -p 8000:8000 \
  -e AZURE_TENANT_ID=... \
  -e AZURE_CLIENT_ID=... \
  -e AZURE_CLIENT_SECRET=... \
  azure-fastmcp-server

FastMCP Configuration

This repository includes a fastmcp.json file used by FastMCP Cloud:

{
  "$schema": "https://gofastmcp.com/public/schemas/fastmcp.json/v1.json",
  "source": {
    "type": "filesystem",
    "path": "server.py",
    "entrypoint": "mcp"
  },
  "environment": {
    "type": "uv",
    "python": ">=3.10",
    "dependencies": [
      "fastmcp==2.14.0",
      "uvicorn[standard]>=0.30.0"
    ]
  },
  "deployment": {
    "transport": "http",
    "host": "0.0.0.0",
    "port": 8000,
    "log_level": "INFO",
    "env": {
      "ENVIRONMENT": "production",
      "FASTMCP_LOG_LEVEL": "INFO"
    }
  }
}

Deploying to FastMCP Cloud

FastMCP Cloud allows you to deploy this server without managing infrastructure.

📘 Official deployment docs: https://gofastmcp.com/deployment/fastmcp-cloud

Security Notes

• No user credentials are accepted
• Service Principal credentials are required
• Runs as a non-root user in Docker
• Health endpoint contains no sensitive data

Versioning

• FastMCP: 2.14.0
• Server Version: 1.0.0