Azure Resource Manager MCP server

This is a repository to track features, bugs, and suggestions for the Azure Resource Manager MCP server.

Overview

Azure Resource Manager (ARM) is the control plane of Azure. Every operation that creates, updates, or deletes Azure resources flows through ARM, regardless of whether it comes from the Azure portal, CLI, PowerShell, REST APIs, or SDKs.

AI agents introduce a new kind of Azure client that needs a consistent, centralized way to interact with ARM. The Azure Resource Manager MCP server is designed to provide that access layer through the Model Context Protocol (MCP), the standard protocol agents use to perform actions and retrieve dynamic data.

Today, the Azure Resource Manager MCP server equips agents with tools to query Azure resources, deploy and manage ARM templates, create or update resources, inspect resource type schemas, and analyze Azure costs and pricing. Its core purpose is to enable AI agents to interact with Azure resources seamlessly, just like other ARM clients. More capabilities will be added in the future.

Features

The Azure Resource Manager MCP server provides the following key features:

• Generate ARG queries dynamically based on user or agent input.
• Validate ARG queries for correctness and security.
• Execute ARG queries against your Azure environment.
• Preview, create, monitor, and cancel ARM template deployments.
• Create or update Azure resources and resource groups.
• Discover Azure resource types, API versions, and schemas.
• Query Azure costs, AKS costs, retail prices, and negotiated pricesheets.

The following tools and schemas are currently registered by the Azure Resource Manager MCP server. Fields marked optional can be omitted.

Tool	Input	Output	Description
create_deployment	request.targetScope: subscriptionId, resourceGroupName, deploymentName; request.definition.properties: mode (Incremental), ARM template, optional parameters, optional validationLevel (Template, Provider, or ProviderNoRbac)	ARM deployment initiation response; monitor it with get_deployment_status	Deploys an ARM template to an Azure resource group.
whatif_deployment	Same target scope and deployment definition as create_deployment; optional whatIfSettings.resultFormat (ResourceIdOnly or FullResourcePayloads)	Predicted resource changes; long-running requests can return locationHeaderUrl for polling	Previews the changes an ARM template deployment would make.
get_deployment_status	request: subscriptionId, resourceGroupName, deploymentName	Current deployment status, result, and any deployment errors	Gets the current status and result of an ARM deployment.
cancel_deployment	request: subscriptionId, resourceGroupName, deploymentName	ARM cancellation response	Cancels an ARM deployment that is currently running.
get_async_operation_status	locationHeaderUrl from an asynchronous resource or what-if response	202 while in progress; terminal 200 response with the operation result	Checks the status of a long-running ARM resource operation. Do not use it for create_deployment; use get_deployment_status.
create_or_update_resource	request.scope: scope type plus identifiers required by that scope; providerNamespace, ordered resourceSegments (resourceType, resourceName), apiVersion, and resource body	ARM PUT response; asynchronous operations include locationHeaderUrl for polling	Creates or updates a single Azure resource at tenant, management group, subscription, resource group, or resource scope.
create_or_update_resource_group	request.scope.subscriptionId, resourceGroupName, and definition.location; optional definition.tags	Created or updated resource group	Creates or updates an Azure resource group, including its location and tags.
list_resource_types	None	Available Azure resource types and their latest API versions	Lists resource types that can be used in ARM templates and direct ARM calls.
get_resource_type_schema	resourceType, apiVersion	JSON schema containing the resource type's required and optional properties	Gets the JSON schema for a resource type and API version.
generate_query	Natural-language prompt	Generated Azure Resource Graph query	Generates an Azure Resource Graph query from a natural-language request.
validate_query	Azure Resource Graph query	Validation result including isValid, syntax errors, validated steps, invalid and extracted properties, and extracted resource types	Validates an Azure Resource Graph query's syntax and property paths.
execute_query	Azure Resource Graph query; optional options: $top, $skipToken, resultFormat	Matching resource data, row count, executed query, total records, truncation flag, and optional skipToken	Runs an Azure Resource Graph query. Continue with skipToken until no token is returned.
query_costs	Azure scope; optional timeframe or from and to, granularity, groupBy, paired filterDimension and filterValues, metric, sortBy, sortDirection, top	Cost and usage rows, sorted as requested; 100 rows by default and up to 5,000	Queries Azure cost and usage data for a subscription or another supported scope.
query_aks_costs	Subscription scope; optional timeframe or from and to, granularity, groupBy, paired filterDimension and filterValues, metric, sortBy, sortDirection, top (registered range: 1-5,000)	AKS cost rows by requested dimensions; 100 rows by default	Queries AKS costs by cluster, namespace, utilization category, and service category.
get_retail_prices	Optional serviceName, armSkuName, armRegionName, meterName, priceType, currencyCode	Raw Retail Prices API response with Items, Count, and optional NextPageLink	Looks up public Azure retail prices by service, SKU, region, meter, or price type.
start_pricesheet_download	Required agreementType (EA, MCA_BillingProfile, or MCA_Invoice) and matching billing scope; billingPeriod (yyyyMM) is required for EA	operationStatusUrl to poll with get_pricesheet_status	Starts an asynchronous EA or MCA Azure pricesheet download.
get_pricesheet_status	operationStatusUrl returned by start_pricesheet_download	InProgress with retryAfterSeconds, or Succeeded with a time-limited downloadUrl	Checks a pricesheet download operation. The server returns the URL but does not download the file.

Cost Management & Pricing tools

Several Cost Management & Pricing tools are enabled by default and need no extra configuration — query_costs, query_aks_costs, get_retail_prices, start_pricesheet_download, and get_pricesheet_status. They let agents answer questions like "What did I spend on Azure this month, broken down by service?" or "How much would a Standard_E4s_v5 VM cost in East US?"

The remaining Cost Management tools — cost forecasting, dimensions, budgets, alerts, and reservation/savings-plan insights — ship as an optional Cost Management toolset you can turn on per client with the x-mcp-toolset: CostManagement header.

See Cost Management & Pricing tools for the full tool list, setup, and example prompts.

Supported Clients

During this preview, Azure Resource Manager MCP server can only be used with a set of MCP Clients. Right now you can use:

• GitHub Copilot Chat in VS Code.
• GitHub Copilot CLI

Support for additional clients will be added based on user feedback and demand. If you have a specific client you'd like to see supported, please open an issue to let us know!

Other client support

For 3rd-party client setup details (app registration, manifest permissions, redirect URI, service principal creation, and connector configuration), see Other client support.

Getting Started

Prerequisites

• VS Code installed
• Valid Azure Account
• GitHub Copilot account

Installation

1. Installing the MCP Server

1. Open: https://aka.ms/JoinARMMCP. VS Code will launch automatically.

2. When prompted inside VS Code, click Install under Azure Resource Manager MCP server to add it to your MCP server configuration.

3. You will be prompted to sign in with your Azure credentials.

2. Open the Chat Interface

1. In VS Code, go to View > Chat, or click the chat icon to the right of the center toolbar header.
2. In the Chat window, click the Configure Tools icon.
3. Ensure Azure Resource Manager MCP server is checked. You can click the dropdown icon to see the available tools under it:
   • create_deployment
   • whatif_deployment
   • get_deployment_status
   • cancel_deployment
   • get_async_operation_status
   • create_or_update_resource
   • create_or_update_resource_group
   • list_resource_types
   • get_resource_type_schema
   • generate_query
   • validate_query
   • execute_query
   • query_costs
   • query_aks_costs
   • get_retail_prices
   • start_pricesheet_download
   • get_pricesheet_status

Usage

1. Open the Chat interface in VS Code.
2. Ensure the Azure Resource Manager MCP server tools are enabled.
3. Type your query or request in natural language.

Demo using the ARG tools

Demo using the Deployment tools

Troubleshooting & FAQ

See the Troubleshooting Guide for common issues and solutions when using the Azure Resource Manager MCP server. For additional questions and answers, refer to the FAQ.

Governance

The Azure Resource Manager MCP server uses the same authentication and authorization context as the signed-in user in VS Code. This means that all tool calls are made on behalf of that user and are subject to the same permissions and access controls defined in Azure.

Blocking Template requests

To prevent any deployments from being made via the ARM MCP server you can apply an Azure Policy to the relevant scope (subscription or resource group) that denies requests made by the deployment tool (create_deployment). To see an example of such a policy, please refer to this sample policy. You will need to specifically block the AppID of the MCP server, which is 22bfbae3-f4e7-485f-be43-8cee15065084.

Contributing

This repository is dedicated solely to gathering feedback from users and contributors. While contributions aimed at clarifying wording or improving documentation details are welcome, the primary purpose of this repository is to facilitate feedback through the creation of issues. Please use issues to share your thoughts, suggestions, or concerns, as this helps us better understand and address your problems!

Code of Conduct

This project has adopted the Microsoft Open Source Code of Conduct. For more information, see the CODE_OF_CONDUCT.md file.

Transparency FAQ

For more information about how the Azure Resource Manager MCP server AI capabilities works, its limitations, and best practices for use, please refer to our Transparency FAQ.

License

This project is licensed under the MIT License. See LICENSE for details.

Data Collection

The Azure Resource Manager MCP server does not collect any personally identifiable information. We collect minimal operational telemetry to capture amount of times a tool is called and service uptime. Our privacy statement is located at https://go.microsoft.com/fwlink/?LinkID=824704.

Trademarks

This project may contain trademarks or logos for projects, products, or services. Authorized use of Microsoft trademarks or logos is subject to and must follow Microsoft’s Trademark & Brand Guidelines. Use of Microsoft trademarks or logos in modified versions of this project must not cause confusion or imply Microsoft sponsorship. Any use of third-party trademarks or logos are subject to those third-party’s policies.