All Azure MCP tools in a single server, implementing the MCP specification to connect AI agents with Azure services. The old Azure/azure-mcp repo was archived on August 25, 2025 and development moved to microsoft/mcp. It can be used alone or with the GitHub Copilot for Azure extension.
What it does
Where Azure MCP Server runs and what it adds, from its documentation
Exposes tools across many Azure services in one server
Microsoft Foundry: list and deploy models, create and evaluate agents
Azure Advisor: list, summarize and act on recommendations
Can be self-hosted remotely on Azure Container Apps
Supports sovereign cloud configuration
When to use it
Good fit
Managing or querying Azure resources from an AI agent
Using Azure from Claude Code via the Azure plugin
Not for
Unreviewed autonomous use: README warns of destructive actions under broad RBAC
How to install Azure MCP Server
You need: Azure account and RBAC permissions; Node.js, .NET 10 SDK, Python or Docker depending on install method
Claude Code: /plugin install azure@claude-plugins-official
Or run npx -y @azure/mcp@latest server start via your client config
Or use .NET 10 dnx Azure.Mcp, PyPI, or Docker per the README
Sign in to Azure so tools use your RBAC permissions
Refresh the tools list and try: List my Azure Storage containers
Permissions and security
What it can access
Operations run with the user's Azure RBAC permissions
Security notes from the maintainers
README: autonomous or misconfigured clients may perform destructive actions; apply least-privilege RBAC
Software may collect usage data sent to Microsoft; telemetry can be turned off
Review the source before granting an agent access to credentials or production systems. We document what the maintainers state; we do not audit code.
Limitations
Safeguard flags are not standardized in MCP and may not be supported by all clients