Azure MCP Server – a Model Context Protocol (MCP) server that lets AI agents interact with many Azure services from VS Code or any MCP-compatible client.
https://github.com/Azure/azure-mcpYou know the routine: check your storage containers, query some logs, manage a few resources. But instead of clicking through the Azure portal or remembering CLI syntax, what if you could just ask your AI agent to do it? Azure MCP Server makes this a reality by connecting any MCP-compatible AI agent directly to your Azure infrastructure.
Every time you need to check an Azure resource, you're breaking flow. Open a browser tab, navigate the portal, find the right service, click through the interface. Or remember the exact az CLI syntax and parameters. Azure MCP Server eliminates this friction by putting Azure operations directly in your chat interface.
Instead of this workflow:
You get this:
This isn't a read-only dashboard – Azure MCP Server handles actual operations across 15+ Azure services. Your AI agent can query Cosmos DB, analyze Log Analytics data, manage Key Vault secrets, and execute KQL queries against your Azure Data Explorer clusters.
Some concrete examples of what you can do:
Azure MCP Server uses the same DefaultAzureCredential flow you're already familiar with – if you can run az login or use Azure CLI, you're ready to go. No additional authentication setup, no new credentials to manage. It automatically uses your existing Visual Studio, Azure CLI, or Azure PowerShell authentication.
The server respects your RBAC permissions too. If you can read a storage account through the portal, you can query it through the MCP server. Same permissions, different interface.
VS Code Integration (Recommended): One-click install buttons that set up everything automatically. Works with both stable and Insiders versions of VS Code, integrating directly with GitHub Copilot.
Standalone Deployment: Run as a Docker container or deploy to Azure App Service for team access. The HTTP API lets any MCP-compatible client connect, not just VS Code.
Modular Service Selection: Install tools for specific Azure services only. Need just Cosmos DB operations? Install the cosmos module. Working with storage and monitoring? Install just those components.
This isn't a demo tool – it's designed for real Azure workloads. The server handles enterprise authentication scenarios, supports production credentials like Managed Identity, and includes comprehensive logging and troubleshooting guides.
The modular architecture means you can deploy exactly what your team needs. Install the full suite for comprehensive Azure access, or deploy specific service modules to different environments based on team responsibilities.
The fastest path is VS Code with the one-click installers, but here's the manual setup if you prefer more control:
{
"servers": {
"Azure MCP Server": {
"command": "npx",
"args": ["-y", "@azure/mcp@latest", "server", "start"]
}
}
}
The server supports both standard IO and SSE transports, so it works with any MCP-compatible client, not just VS Code integrations.
Azure MCP Server turns your AI agent into an Azure operations assistant that understands your infrastructure context. Stop juggling browser tabs and CLI commands – let your agent handle the Azure operations while you focus on building.