Posted in

Azure Resource Manager MCP Server

🔹 Feature: Azure Resource Manager MCP Server — Public Preview – Finally
🔹 What It Does: A remote MCP server that gives AI agents first-class access to Azure infrastructure operations through Azure Resource Manager. Generate, validate, and execute Azure Resource Graph queries from natural language, deploy and monitor ARM templates, and let your AI agent reason across your entire Azure estate — all secured by your existing RBAC and Azure Policy. 🤖

💡 Thesis: AI agents can now operate Azure infrastructure natively — no more screenshots, manual KQL, or brittle scripts.

What Is It Giving You:

✅ Natural Language → Azure Resource Graph: Stop writing KQL from scratch. Tell your agent “Find all VMs in my subscription without managed disks” or “Show me resources created in the last 30 days without required tags” — the server generates the ARG query, validates it, and executes it across every Azure resource type.
✅ Full ARM Template Lifecycle: Pass an ARM template, kick off deployment at resource group scope, check deployment status, and cancel in-flight deployments if something looks wrong. AI-driven infra deployment with proper guardrails.
✅ Remote MCP — Not Yet Another Local Server: Unlike most MCP servers you spin up on localhost, this one runs as a remote service. Less setup, no version drift across team members, no “works on my machine” agent behavior.
✅ Cross-Resource Reasoning: ARG queries span every Azure resource type in your scope. Compliance checks, tagging audits, drift detection, cost hygiene — all become natural-language questions instead of custom scripts.
✅ Security Model You Can Actually Defend: Every operation runs in the context of the signed-in user. No service principals with god-mode. Want to block AI deployments entirely? Apply an Azure Policy to deny the createtemplatedeployment tool at subscription or RG scope. Done.
✅ Custom Agent Building Block: Wire it into GitHub Copilot custom agents to automate compliance scans, drift detection, tag enforcement, or change tracking. The MCP server is the tool — your agent is the logic.
✅ Open Feedback Loop: Currently works with GitHub Copilot Chat in VS Code. More clients coming based on demand — vote with GitHub issues.

🚀 The shift: The era of “screenshot the Azure portal and ask Copilot what’s wrong” is ending. AI agents now get a real, governed handle on ARM itself.

🌐 https://github.com/Azure/Azure-Resource-Manager-MCP/blob/main/README.md

Microsoft Certified Trainer, Office 365, AWS, Azure and Cloud Expert-Architect. In the IT world for over than 20 years.

Apart from the main area of Microsoft Azure expert in the field of infrastructure servers Windows Server 2003-2019, Microsoft Active Directory, Hyper-V Private Cloud, IIS, System Center, SQL.

Private Cloud, System Center, Hyper-V, Open Stack Expert and all Microsoft products Expert. Linux Server administrator.

My Azure community projects:

https://mazeball.azurewebsites.net/
https://github.com/MariuszFerdyn?tab=repositories

More