Part 1 of the Azure Skills Plugin series

---

Coding agents like GitHub Copilot and Claude Code are great at code, but getting your app to production on Azure is not just about writing code.

Really, it is about making the right calls. Which service fits this app? Which SKU fits this workload? Should this be App Service, Container Apps, Functions, AKS, or something else entirely? What needs to be validated before deploy? Which permissions, quotas, and guardrails matter?

That is exactly why skills are taking off: they give agents practical knowledge on demand instead of forcing them to guess.

Want to try this right now instead of reading the whole post first?

aka.ms/azure-skills

The Azure Skills Plugin brings that idea to Azure work. It bundles curated Azure skills, the Azure MCP Server, and the Foundry MCP Server in one install so a coding agent can do more than explain Azure. It can reason about Azure workflows and then execute them with real tools.

GitHub Copilot in VS Code is an obvious place to use the Azure skills plugin. So is Copilot CLI. But the bigger story is portability: because this is built around open skills and plugin patterns, the same package can also light up experiences in Claude Code and other tools that support plugins and skills.

Why This Hits Different

This is not another prompt pack.

It is not a folder full of clever one-liners. It is a carefully curated set of detailed Azure instructions:

• Skills package Azure expertise as reusable workflows, decision trees, and guardrails.
• MCP servers give the agent tools to act on Azure and Foundry.
• The plugin ships both together so the guidance and the execution layer stay aligned.

If you’ve been following the broader agent world, this pattern should feel familiar. Skills are becoming the way teams turn tribal knowledge into something reusable. The Azure Skills Plugin follows that pattern and applies it to real Azure work across modern coding agents.

Works Across Agent Hosts

The Azure Skills Plugin is super useful in any one host on its own. But what makes it especially powerful is that the same Azure knowledge and tooling can travel across hosts.

• GitHub Copilot in VS Code gives you the in-editor experience most developers will see first.
• Copilot CLI gives you the same skills in a terminal-first workflow that is becoming very popular.
• Claude Code can use the same plugin and skills for Azure-capable coding workflows.
• Other tools that support skills and plugins can benefit from the same packaged Azure expertise.

Developers already have their preferred tools and workflows, and Azure should meet them where they already build.

One Install, Three Layers of Capability

19+ Azure Skills – The Brain

Skills teach the agent how Azure work gets done.

They do not just list tools, but instead tell the agent when to use them, in what order, and what to avoid. The skills capture how an Azure expert would approach the job:

• azure-prepare analyzes your project and generates infrastructure code, Dockerfiles, and azure.yaml
• azure-validate runs pre-flight checks before you burn time on a failed deployment
• azure-deploy orchestrates the actual deployment pipeline via azd
• azure-cost-optimization finds waste and produces concrete savings recommendations
• azure-diagnostics troubleshoots real failures with logs, metrics, and KQL

There are 19+ skills covering compute, observability, AI, compliance, storage, migration, RBAC, messaging, and more.

Azure MCP Server – The Hands

The Azure MCP (Model Context Protocol) Server provides over 200 structured tools across 40+ Azure services.

Here are the actual operations the agent can perform: listing resources, checking prices, querying logs, running diagnostics, provisioning infrastructure, and driving deployment workflows.

If you have seen my earlier Azure MCP content, you already know how powerful this tools layer is. The plugin brings the missing orchestration layer on top of it.

Foundry MCP Server – The AI Specialist

The Foundry MCP Server connects the agent to Microsoft Foundry for model deployment, agent management, and model catalog workflows.

If you are building AI-powered apps on Azure, this is the part that lights up model and agent scenarios instead of stopping at generic cloud guidance.

See the Flow

That is the core value proposition:

• Skills guide.
• MCP executes.
• The plugin packages both.

Before and After

Here is what the same prompt looks like with and without the plugin.

Scenario	Without the plugin	With the plugin
Prompt	“Deploy my Python Flask API to Azure”	“Deploy my Python Flask API to Azure”
Copilot behavior	Generic tutorial, maybe some az commands, maybe a docs link	Activates azure-prepare, azure-validate, and azure-deploy in sequence
Who makes the decisions?	You	Copilot follows an Azure workflow encoded by the skills
What gets produced?	Advice	Dockerfile, infrastructure files, azure.yaml, validation output, deployment actions
What actually runs?	Usually nothing	Real MCP tool calls against Azure and Foundry

Same prompt. Completely different outcome.

Without the plugin, Copilot gives you directions.

But with the plugin, Copilot gets an Azure playbook and the tools to follow it.

Why Skills Matter Right Now

Skills solve a very specific problem: agents are smart, but they are often missing the situational knowledge required to do real work reliably.

That is why the skills model is resonating:

• They load on demand. You can install a large capability set without dragging every workflow into every prompt.
• They are auditable. Skills are plain text, version-controlled, reviewable, and easy to improve.
• They scale expertise. One team can package hard-won Azure knowledge once and reuse it over and over.
• They pair naturally with MCP. Skills provide the workflow. MCP provides the execution layer.

The Azure Skills Plugin matters because it turns that general idea into something concrete for developers shipping on Azure today.

Real Tools, Real Guardrails

The plugin uses real tools against real Azure resources. That is the point. It also means the right model is not blind trust. The right model is curated skills, trusted plugin sources, tool approvals, and least-privilege access.

That is a strength, not a weakness. Ultimately, the whole value of this approach is that Copilot is grounded in real workflows and real operations instead of hallucinated cloud advice.

What’s in the Box

When you install the Azure Skills Plugin, you get:

Component	What It Does	How Many
Azure Skills	Azure expertise, workflows, and guardrails	19+ skills
Azure MCP Server	Azure service tools	~200 tools, 40+ services
Foundry MCP Server	AI model and agent workflows	AI Foundry integration

Skills install into .github/plugins/azure-skills in your workspace. The MCP servers are configured via a .mcp.json file that the plugin sets up for you.

Who Is This For?

• Startups and AI founders that want to focus on shipping solutions, models, agents, and user experience instead of building and managing cloud infrastructure by hand
• Agent builders who want Azure + Foundry workflows in the same toolchain
• Developers who want to stop bouncing between docs, portal tabs, and complicated CLI commands
• Cloud engineers who want a faster path from request to execution
• DevOps teams who want repeatable infrastructure-as-code instead of ad hoc chat answers
• Anyone who has ever stared at Azure’s service catalog and thought, “which one do I actually need?”

You do not need to be an Azure expert. That is the whole point. The expertise comes with the plugin.

Where to Get It

Install the Azure Skills Plugin: https://aka.ms/azure-plugin

One link. One install. A much smarter Azure experience across the coding agents that support it.

Get Started in 60 Seconds

If you want the short version, here it is.

Prerequisites

• A compatible host such as GitHub Copilot in VS Code, Copilot CLI, or Claude Code
• Node.js 18+ on your PATH (the MCP servers use npx)
• An Azure account if you want to do real Azure operations
• Azure CLI (az) installed and signed in with az login
• Azure Developer CLI (azd) installed and signed in with azd auth login if you want to use deployment workflows

Install

1. Go to aka.ms/azure-plugin
2. Follow the install flow for your host
3. Let it add the Azure skills plus the Azure and Foundry MCP server configuration

For host-specific install commands and troubleshooting, see the Azure Skills Plugin README.

Verify

After install, confirm three things:

• The Azure skills package is available
• The Azure MCP Server is configured and running
• The Foundry MCP Server is configured and running

In VS Code-style hosts, this usually means you should see skills available to the agent and a .mcp.json with azure and foundry configured.

Smoke Test

Try one prompt in your host of choice:

“What Azure services would I need to deploy this project?”

Then try one tool-backed prompt:

“List my Azure resource groups”

If the first prompt gives you structured Azure guidance and the second returns real resources from your subscription, you are up and running.

What’s Next in This Series

This post is the kickoff. Here is what is coming next:

1. Install Guide – Step-by-step setup with screenshots
2. How Skills + MCP Work Together – The architecture under the hood
3. The Core Workflow – Prepare → validate → deploy
4. Skills That Save You Hours – Cost optimization, diagnostics, compute, and more
5. Inside a SKILL.md – How skills actually work
6. Real-World Walkthrough – Full-stack app to production, entirely through Copilot
7. What’s Next – Where Azure skills, agents, and MCP go from here

---

Try It Yourself

1. Install the plugin: aka.ms/azure-plugin
2. Open your preferred compatible host with a project you would like to deploy to Azure
3. Try: “What Azure services would I need to deploy this project?”
4. Then try: “List my Azure resource groups”
5. Then try: “Prepare this app for Azure”
6. Watch the skills kick in

If you have seen Azure MCP before, this is the next step: the same hands, now with a brain and a plan.

---

Next up: How to Install the Azure Skills Plugin