Azure MCP Server: Let AI Manage Resources, Query Costs, and Monitor Services

Why Create an Azure MCP Server?

Azure hosts production workloads for millions of teams. Checking resource status, reviewing cost reports, scaling VMs, or debugging service health means switching to the Azure Portal, navigating through blades, running queries in Cost Management or Monitor.

With an MCP server connected to Azure, AI agents can answer infrastructure questions directly from your editor. Check costs, list resources, and monitor services without opening a browser tab.

What AI Agents Can Do With Azure Access

Once connected, Claude, Cursor, or Windsurf can:

• List resources — "What VMs are running in the production resource group?"
• Check costs — "How much did we spend on Azure last month?"
• Monitor health — "Are there any service health incidents right now?"
• Query metrics — "What's the CPU utilization of the API server VM?"
• Review deployments — "What was the last deployment to the staging slot?"
• Check quotas — "How close are we to the vCPU limit in East US?"

How It Works

1. Open DataFaucet and enter https://portal.azure.com
2. Log in to your Azure account in the cloud browser
3. Navigate through the portal — check resources, view cost reports, open Monitor
4. DataFaucet captures each interaction as an MCP tool
5. Deploy — your AI agents now have structured access to Azure

Every portal action you perform gets turned into a callable tool. Cost queries become a "check_costs" tool. Resource listings become "list_resources". No API keys or service principals needed.

Example Queries

"What resources are in the production resource group?"
"Show this month's Azure spending broken down by service"
"Are all VMs in the web tier running?"
"What alerts fired in the last 24 hours?"
"What's the current DTU usage on the SQL database?"
"List all storage accounts and their sizes"

When This Saves Time

• Cost reviews — ask about spending instead of navigating Cost Management blades
• Incident response — check service health and VM status from terminal during outages
• Architecture planning — query existing resource layouts while designing new services
• Compliance audits — list resources by tag, region, or configuration
• Sprint planning — review infrastructure state while estimating DevOps work

Comparison: Azure CLI vs MCP Server

The Azure CLI (az) gives you programmatic access but requires remembering commands, flags, and resource IDs. An MCP server lets you ask questions in natural language. The AI agent handles the translation.

Both work. The MCP server reduces context-switching when you're mid-task in your editor.

Security Considerations

Azure Portal access uses your browser session credentials. The MCP server inherits whatever permissions your logged-in account has. For production use:

• Use a read-only Azure role (Reader) for the service principal or account
• Scope access to specific resource groups
• DataFaucet captures interactions as tools — it doesn't store your Azure credentials
• Rotate the MCP server API key if team members change

Getting Started

Log into the Azure Portal. Paste https://portal.azure.com into DataFaucet. Browse your resources, check costs, view monitoring dashboards. Deploy. Your AI agents can now query Azure infrastructure from your editor.

Create your Azure MCP server →