MCP Server

Cloud InfrastructureMicrosoft AzureMIT License

Azure MCP Server

by Microsoft / Community · azure/mcp-server

Azure MCP Server connects your AI coding assistant directly to Microsoft Azure, exposing Blob Storage, Azure Functions, Cosmos DB, Azure AI services, and resource management as MCP tools. Ask your agent to upload a file to Blob Storage, invoke a serverless Function, query a Cosmos DB collection, or call an Azure OpenAI model — all without touching the Azure Portal or writing boilerplate SDK code.

Azure MCP Server uses the Azure Identity SDK's DefaultAzureCredential for authentication, which automatically discovers your Azure CLI session, service principal, or managed identity. This means developers already using the Azure CLI get zero-configuration access, while production environments use managed identities or service principals with precise RBAC roles for security compliance.

Services Covered

Blob, Functions, Cosmos...

Azure AD

Auth

DefaultAzureCredential

<5min

Setup

npm install + az login

MIT

License

open source

Quick Install

claude mcp add azure -- npx @azure/mcp-server

Key Features

Azure Blob Storage

List containers and blobs, upload files with configurable content types and metadata, download blobs, delete objects, and generate time-limited SAS tokens for secure external sharing. The agent handles chunked uploads for large files and automatically paginates list responses for containers with thousands of blobs.

Azure Functions

Invoke HTTP-triggered Azure Functions, deploy new function code packages, check function status, retrieve execution logs from Application Insights, and manage function app settings. Supports both consumption and premium plan functions in any supported runtime (Node.js, Python, .NET, Java).

Cosmos DB Operations

Query Cosmos DB containers using SQL API syntax, get and upsert documents, list databases and containers, and retrieve throughput configuration. The agent constructs correct Cosmos DB query syntax automatically, handles continuation tokens for paginated results, and surfaces partition key errors with clear explanations.

Azure AI Services

Call Azure OpenAI chat completions and embeddings endpoints, analyze images with Computer Vision, detect language and extract key phrases with Text Analytics, and transcribe audio with Speech Services — all using your Azure subscription's deployed models and endpoints without exposing raw API keys in your code.

Resource Management

List subscriptions, resource groups, and resources by type or tag. Check deployment status, retrieve resource properties, and identify orphaned resources for cost optimization. The agent can generate a complete inventory of your Azure resources across multiple resource groups in a single query.

Azure Monitor & Logs

Query Azure Monitor logs using KQL (Kusto Query Language), retrieve metric time series for any Azure resource, and check alert rule status. The agent helps diagnose production issues by correlating application logs with infrastructure metrics without requiring KQL expertise from the developer.

Execution Brief

Use this page as a rollout checklist, not just reference text.

Suggest update

Tool Mapping Lens

Organize Tools by Workflow Phase

Catalog-oriented pages work best when users can map discovery, evaluation, and rollout in a clear path instead of reading an undifferentiated list.

Define the job-to-be-done first
Group tools by stage
Prioritize by adoption friction

Actionable Utility Module

Skill Implementation Board

Use this board for Azure MCP Server before rollout. Capture inputs, apply one decision rule, execute the checklist, and log outcome.

Input: Objective

Deliver one measurable improvement with azure mcp server claude code microsoft azure blob storage cosmos db functions ai cloud automation

Input: Baseline Window

20-30 minutes

Input: Fallback Window

8-12 minutes

Decision Trigger	Action	Expected Output
Input: one workflow objective and release owner are defined	Run preview execution with fixed acceptance criteria.	Go or hold decision backed by repeatable evidence.
Input: output quality below baseline or retries increase	Limit scope, isolate root issue, and rerun controlled test.	One confirmed correction path before wider rollout.
Input: checks pass for two consecutive replay windows	Promote to broader traffic with fallback path active.	Stable rollout with low operational surprise.

Execution Steps

Record objective, owner, and stop condition.
Execute one controlled preview run.
Measure quality, latency, and correction burden.
Promote only when pass criteria are stable.

Output Template

tool=azure mcp server claude code microsoft azure blob storage cosmos db functions ai cloud automation
objective=
preview_result=pass|fail
primary_metric=
next_step=rollout|patch|hold

Share execution feedback

What Is Azure MCP Server?

Azure MCP Server is a Model Context Protocol bridge that exposes Microsoft Azure service APIs as structured tools your AI coding assistant can invoke conversationally. It covers Blob Storage, Azure Functions, Cosmos DB, Azure AI services, and Azure Resource Manager, translating natural language requests into precise Azure REST API calls and returning formatted results to the conversation.

Managing Azure infrastructure typically involves juggling the Azure Portal, CLI, ARM templates, and service-specific SDKs. Azure MCP Server collapses this into a single conversational interface inside your IDE. Instead of navigating menus or memorizing CLI commands, you ask: "list all Function Apps in my production resource group and show me the last error in their Application Insights logs" — and the agent does the rest.

Authentication uses the Azure Identity SDK's DefaultAzureCredential, which automatically discovers credentials from the Azure CLI session, environment variables for service principals, or managed identities in Azure-hosted environments. This means zero reconfiguration between local development (where you use az login) and production deployments (where managed identity takes over).

Azure MCP Server integrates with the broader MCP ecosystem to enable end-to-end workflows. Combine it with GitHub MCP to trigger Azure Function deployments on PR merges, Filesystem MCP to sync local build output to Blob Storage, or Postgres MCP to migrate data between Azure SQL and Cosmos DB. The agent orchestrates cross-service operations that would otherwise require custom CI/CD scripts.

How to Calculate Better Results with azure mcp server claude code microsoft azure blob storage cosmos db functions ai cloud automation

Install the Azure MCP server package using the quick install command above. The package uses the Azure Identity SDK and Azure SDK for JavaScript, both of which are included as dependencies. Node.js 18 or later is required.

Authenticate by running "az login" in your terminal if you haven't already. Set your target subscription using "az account set --subscription <subscription-id>". The MCP server will automatically use your Azure CLI session credentials when launched. For CI environments, set AZURE_CLIENT_ID, AZURE_CLIENT_SECRET, and AZURE_TENANT_ID environment variables for service principal authentication.

Register the server with Claude Code or your MCP client using the quick install command. For other clients, add the server configuration to your MCP JSON config file. If you need to target a specific Azure environment (such as Azure China or Azure Government), set the AZURE_AUTHORITY_HOST environment variable accordingly.

Verify the setup by asking your agent to list your Azure resource groups or Blob Storage containers. A successful response confirms authentication is working. Test a write operation — such as uploading a small test file to a Blob Storage container — to confirm your service principal or user account has the necessary RBAC permissions for your intended workflows.

Treat this page as a decision map. Build a shortlist fast, then run a focused second pass for security, ownership, and operational fit.

When a team keeps one shared selection rubric, tool adoption speeds up because evaluators stop debating criteria every time a new option appears.

Worked Examples

Cosmos DB data migration and validation

You need to migrate a collection of documents from one Cosmos DB container to another with a schema transformation applied to each document
Ask the agent to query all documents from the source container using Azure MCP, paging through results with the continuation token until all documents are retrieved
The agent applies the required schema transformation to each document — renaming fields, converting data types, and adding new computed properties
Transformed documents are upserted into the destination container in batches using Cosmos DB batch write operations via Azure MCP
The agent queries both containers to compare document counts and runs sample spot-check queries to verify transformed data integrity
A migration summary is generated showing total documents migrated, any documents that failed validation, and the time elapsed for the entire operation

Outcome: A complete data migration with schema transformation and validation completed conversationally, without writing custom migration scripts or manually monitoring progress in the Azure Portal.

Azure Function performance investigation

Users are reporting slow response times from an HTTP-triggered Azure Function during peak hours
Ask the agent to query Azure Monitor metrics for the Function App to retrieve execution count, duration percentiles, and failure rate over the past 24 hours
The agent identifies time windows with elevated P95 execution times and cross-references them with Application Insights dependency tracking data
The agent queries the Function's Application Insights logs using KQL to find the specific operations with the highest duration during the slow periods
The agent retrieves the Function App's current configuration settings — including host.json concurrency settings and connection string configuration — via Azure MCP resource management tools
Based on the metrics and logs, the agent generates a prioritized list of configuration changes to investigate: scaling settings, connection pool size, cold start mitigation options, and dependencies with high latency

Outcome: A complete performance root-cause investigation that surfaces specific configuration changes to test, achieved without opening the Azure Portal or writing any KQL queries manually.

Frequently Asked Questions

What is the Azure MCP Server?

Azure MCP Server is a Model Context Protocol integration that exposes Microsoft Azure service APIs as conversational tools your AI coding assistant can invoke. It covers key Azure services including Blob Storage for object storage, Azure Functions for serverless compute, Cosmos DB for globally distributed databases, and Azure AI services for vision, language, and speech APIs. Instead of switching to the Azure Portal or writing SDK code from scratch, you can ask your agent to upload a blob, deploy a Function, or call an Azure OpenAI endpoint — all from Claude Code or any MCP-compatible client.

How does Azure MCP Server authenticate with my Azure subscription?

Azure MCP Server uses the Azure Identity SDK's DefaultAzureCredential, which automatically tries multiple authentication methods in order: environment variables (AZURE_CLIENT_ID, AZURE_CLIENT_SECRET, AZURE_TENANT_ID for service principal auth), managed identity (when running on Azure VMs or App Service), Azure CLI credentials, and Visual Studio Code Azure extension credentials. For local development, the simplest approach is to run "az login" with the Azure CLI and set your target subscription with "az account set --subscription <id>". No additional configuration is needed after that.

Which Azure services does Azure MCP support?

Azure MCP Server supports the most commonly used Azure services: Blob Storage for container and blob operations (list, upload, download, delete, generate SAS tokens), Azure Functions for invocation and deployment, Cosmos DB for document CRUD operations and queries, Azure AI Services including Azure OpenAI, Computer Vision, and Text Analytics, and Azure Resource Manager for subscription and resource group listing. The exact set of supported operations varies by implementation — check the package README for the current service coverage.

Can Azure MCP Server work with Azure OpenAI?

Yes. Azure MCP Server includes integration with Azure OpenAI Service, allowing the agent to call your deployed Azure OpenAI models (GPT-4, GPT-4o, embeddings) using your subscription endpoint and API key. This is particularly useful for testing prompt templates, evaluating model responses, or building chained workflows where the coding agent uses Azure OpenAI to process data that was retrieved from Cosmos DB or Blob Storage in the same conversation.

How does Azure MCP Server compare to AWS MCP Server?

Azure MCP Server and AWS MCP Server are functionally equivalent in concept — both bridge AI agents to cloud provider APIs via MCP. The choice depends on which cloud provider your infrastructure uses. Azure MCP offers native integration with Azure-specific services like Cosmos DB, Azure Active Directory, and the Azure AI Foundry model catalog. AWS MCP integrates with services like DynamoDB, S3, and AWS Bedrock. If your organization uses both clouds, you can run both MCP servers simultaneously in your MCP client configuration and the agent will select the appropriate one based on context.

Is Azure MCP Server suitable for enterprise environments?

Yes. Azure MCP Server supports enterprise-grade authentication via service principals and managed identities, making it compatible with corporate Azure AD tenants and RBAC-controlled subscriptions. Assign a dedicated service principal with a minimum-privilege role (such as Storage Blob Data Reader or Cosmos DB Account Reader) to limit what the agent can access. All Azure API calls are logged in Azure Monitor and Activity Log, providing full auditability of agent actions in your Azure environment.

Missing a better tool match?

Send the exact workflow you are solving and we will prioritize a new comparison or rollout guide.

Submit feedback