Azure MCP Server – Cloud Resource Management
Microsoft's official Azure MCP Server gives AI agents structured access to Azure services and resource-management workflows. Use it to inspect, manage, deploy, query, and troubleshoot supported Azure resources directly from MCP-compatible development tools.
Overview
Microsoft's official Azure MCP Server connects compatible AI agents to Azure
services through the Model Context Protocol. It provides a unified local server
that exposes tools for working with supported Azure resources, allowing an
agent to perform cloud-development and operations tasks from an IDE or other
MCP client.
What the MCP server enables
Azure MCP groups tools by Azure service namespace. Depending on the installed
version, enabled namespaces, authenticated identity, and Azure role assignments,
an AI agent can:
- Discover subscriptions, resource groups, and supported Azure resources.
- Inspect and manage storage, databases, networking, compute, and application services.
- Query logs, metrics, and other operational data.
- Deploy or update supported Azure resources and applications.
- Retrieve service-specific configuration and troubleshooting information.
- Coordinate workflows that span multiple Azure services.
- Limit exposed tools to selected service namespaces for focused workflows.
When to use it
Use Azure MCP when an AI workflow needs live Azure context or must perform
authorized Azure operations. Typical examples include listing storage accounts,
inspecting resource groups, querying application logs, troubleshooting a cloud
deployment, reviewing configuration, provisioning development resources, or
automating repetitive Azure management tasks from an AI coding assistant.
Connection and authentication
The official server runs locally over stdio with the
@azure/mcp@latest npm package. The documented command is
npx -y @azure/mcp@latest server start.
Azure MCP authenticates through Microsoft Entra ID and can discover credentials
from supported local developer tools such as Azure CLI, Azure Developer CLI,
Visual Studio Code, Visual Studio, and Azure PowerShell. It can also use
environment-based credentials and interactive browser authentication through
the Azure Identity credential chain. The authenticated identity must have the
Azure RBAC permissions required by each requested operation.
Key considerations
Tool access does not bypass Azure authorization. Every action is limited by the
signed-in identity, subscription access, resource permissions, policies, and
service-specific restrictions. Use least-privilege Azure roles, review write or
destructive actions before approval, and restrict the server to only the
namespaces needed for the workflow. The older Azure/azure-mcp repository is
archived; current development is maintained in Microsoft's official MCP
repository. Keep the npm package updated and avoid exposing production
subscriptions to untrusted clients.
Supported Transports
stdio
Command: npx
Args:
-y@azure/mcp@latestserverstart
Frequently Asked Questions
- When should an AI agent use the Azure MCP Server?
- Use it when a workflow needs live Azure resource context or authorized cloud operations, such as discovering resources, querying logs, inspecting configuration, deploying applications, troubleshooting services, or managing development infrastructure.
- What does the Azure MCP Server add to an AI agent's capabilities?
- It gives the agent structured access to supported Azure services and resource-management tools, allowing it to work with current Azure state instead of relying only on static model knowledge or manually copied portal information.
- What can an AI agent access or manage through Azure MCP?
- Depending on enabled namespaces and Azure permissions, the agent can work with subscriptions, resource groups, storage, databases, networking, compute, application services, logs, metrics, deployments, and other supported Azure resources.
- How is authentication configured for the Azure MCP Server?
- The server uses Microsoft Entra ID through the Azure Identity credential chain. It can discover credentials from Azure CLI, Azure Developer CLI, Visual Studio Code, Visual Studio, Azure PowerShell, environment variables, or interactive browser sign-in. Azure RBAC still controls every operation.
- Which transport should be used for the Azure MCP Server?
- Use stdio with the official @azure/mcp npm package launched through npx. Microsoft's documented configuration runs the server locally, and no hosted Streamable HTTP or SSE transport is included in this YAML.