Generated banner for MCP Builder
Generated logo for MCP Builder

MCP Builder

MicrosoftOfficialPublished by Microsoft

  • Skill
  • Guidelines
  • Software development
  • MIT
  • Low risk
  • Has scripts

From the author

Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP), Node/TypeScript (MCP SDK), or C#/.NET (Microsoft MCP SDK).
— the description by Microsoft

AutoGineer summary

Written by AutoGineer from the skill’s own files, not by the author

mcp-builder provides a structured, four-phase workflow for developing MCP servers that enable LLMs to interact with external services. It covers the entire lifecycle from initial research and planning, through implementation in TypeScript, Python, or C#, to testing and creating robust evaluations. The guide emphasizes best practices for tool design, schema definition, and error handling to ensure agents can effectively utilize the exposed functionality.

What it does

  • Provides a step-by-step process for building MCP servers.
  • Offers guidance on selecting the right language SDK (TypeScript, Python, C#).
  • Defines standards for tool naming, schema design, and context management.
  • Includes instructions for implementing core infrastructure like authentication and pagination.
  • Outlines a methodology for creating XML-based evaluations to test agent performance.
  • Provides references for Microsoft-specific MCP patterns and Azure integration.

When to use it

  • When you need to integrate custom internal APIs or third-party SaaS into an LLM agent.
  • When you are building a new MCP server and want to follow industry best practices.
  • When you need to evaluate the effectiveness of your MCP server's tools.
  • When deciding between building a custom server or using existing Microsoft MCP infrastructure.

Advantages

  • Standardizes the development process for consistent, high-quality MCP servers.
  • Promotes the use of structured schemas for better agent interaction.
  • Includes a clear path for testing and validating server performance.
  • Supports multiple programming languages and transport mechanisms.

Usage examples

“I need to build an MCP server for my internal CRM API. What is the recommended workflow?”
The agent will guide you through the four-phase process: planning your tool definitions, selecting the appropriate SDK, implementing the server with proper error handling, and creating an evaluation suite.
“How should I structure my tool definitions for an MCP server?”
The agent will provide patterns for clear, action-oriented naming, using Zod or Pydantic for input schemas, and defining output schemas to help agents process data effectively.

Install

  1. Download the skill as a ZIP:

    https://mp-staging.autogineer.com/agent-skills/microsoft/mcp-builder/download
  2. Unzip it into the cross-client skills folder of your project:

    mkdir -p .agents/skills && unzip mcp-builder.zip -d .agents/skills

Documentation by Microsoft

MCP Server Development Guide

Overview

Create MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. The quality of an MCP server is measured by how well it enables LLMs to accomplish real-world tasks.


Microsoft MCP Ecosystem

Microsoft provides extensive MCP infrastructure for Azure and Foundry services. Understanding this ecosystem helps you decide whether to build custom servers or leverage existing ones.

Server Types
TypeTransportUse CaseExample
LocalstdioDesktop apps, single-user, local devAzure MCP Server via NPM/Docker
RemoteStreamable HTTPCloud services, multi-tenant, Agent Servicehttps://mcp.ai.azure.com (Foundry)
Microsoft MCP Servers

Before building a custom server, check if Microsoft already provides one:

ServerTypeDescription
Azure MCPLocal48+ Azure services (Storage, KeyVault, Cosmos, SQL, etc.)
Foundry MCPRemotehttps://mcp.ai.azure.com - Models, deployments, evals, agents
Fabric MCPLocalMicrosoft Fabric APIs, OneLake, item definitions
Playwright MCPLocalBrowser automation and testing
GitHub MCPRemotehttps://api.githubcopilot.com/mcp

Full ecosystem: See 🔷 Microsoft MCP Patterns for complete server catalog and patterns.

When to Use Microsoft vs Custom
ScenarioRecommendation
Azure service integrationUse Azure MCP Server (48 services covered)
AI Foundry agents/evalsUse Foundry MCP remote server
Custom internal APIsBuild custom server (this guide)
Third-party SaaS integrationBuild custom server (this guide)
Extending Azure MCPFollow Microsoft MCP Patterns

Process

🚀 High-Level Workflow

Creating a high-quality MCP server involves four main phases:

Phase 1: Deep Research and Planning
1.1 Understand Modern MCP Design

API Coverage vs. Workflow Tools: Balance comprehensive API endpoint coverage with specialized workflow tools. Workflow tools can be more convenient for specific tasks, while comprehensive coverage gives agents flexibility to compose operations. Performance varies by client—some clients benefit from code execution that combines basic tools, while others work better with higher-level workflows. When uncertain, prioritize comprehensive API coverage.

Tool Naming and Discoverability: Clear, descriptive tool names help agents find the right tools quickly. Use consistent prefixes (e.g., github_create_issue, github_list_repos) and action-oriented naming.

Context Management: Agents benefit from concise tool descriptions and the ability to filter/paginate results. Design tools that return focused, relevant data. Some clients support code execution which can help agents filter and process data efficiently.

Actionable Error Messages: Error messages should guide agents toward solutions with specific suggestions and next steps.

1.2 Study MCP Protocol Documentation

Navigate the MCP specification:

Start with the sitemap to find relevant pages: https://modelcontextprotocol.io/sitemap.xml

Then fetch specific pages with .md suffix for markdown format (e.g., https://modelcontextprotocol.io/specification/draft.md).

Key pages to review:

  • Specification overview and architecture

  • Transport mechanisms (streamable HTTP, stdio)

  • Tool, resource, and prompt definitions

1.3 Study Framework Documentation

Language Selection:

LanguageBest ForSDK
TypeScript (recommended)General MCP servers, broad compatibility@modelcontextprotocol/sdk
PythonData/ML pipelines, FastAPI integrationmcp (FastMCP)
C#/.NETAzure/Microsoft ecosystem, enterpriseMicrosoft.Mcp.Core

Transport Selection:

TransportUse CaseCharacteristics
Streamable HTTPRemote servers, multi-tenant, Agent ServiceStateless, scalable, requires auth
stdioLocal servers, desktop appsSimple, single-user, no network

Load framework documentation:

For TypeScript (recommended):

  • TypeScript SDK: Use WebFetch to load https://raw.githubusercontent.com/modelcontextprotocol/typescript-sdk/main/README.md

  • ⚡ TypeScript Guide - TypeScript patterns and examples

For Python:

  • Python SDK: Use WebFetch to load https://raw.githubusercontent.com/modelcontextprotocol/python-sdk/main/README.md

  • 🐍 Python Guide - Python patterns and examples

For C#/.NET (Microsoft ecosystem):

1.4 Plan Your Implementation

Understand the API: Review the service's API documentation to identify key endpoints, authentication requirements, and data models. Use web search and WebFetch as needed.

Tool Selection: Prioritize comprehensive API coverage. List endpoints to implement, starting with the most common operations.


Phase 2: Implementation
2.1 Set Up Project Structure

See language-specific guides for project setup:

2.2 Implement Core Infrastructure

Create shared utilities:

  • API client with authentication

  • Error handling helpers

  • Response formatting (JSON/Markdown)

  • Pagination support

2.3 Implement Tools

For each tool:

Input Schema:

  • Use Zod (TypeScript) or Pydantic (Python)

  • Include constraints and clear descriptions

  • Add examples in field descriptions

Output Schema:

  • Define outputSchema where possible for structured data

  • Use structuredContent in tool responses (TypeScript SDK feature)

  • Helps clients understand and process tool outputs

Tool Description:

  • Concise summary of functionality

  • Parameter descriptions

  • Return type schema

Implementation:

  • Async/await for I/O operations

  • Proper error handling with actionable messages

  • Support pagination where applicable

  • Return both text content and structured data when using modern SDKs

Annotations:

  • readOnlyHint: true/false

  • destructiveHint: true/false

  • idempotentHint: true/false

  • openWorldHint: true/false


Phase 3: Review and Test
3.1 Code Quality

Review for:

  • No duplicated code (DRY principle)

  • Consistent error handling

  • Full type coverage

  • Clear tool descriptions

3.2 Build and Test

TypeScript:

  • Run npm run build to verify compilation

  • Test with MCP Inspector: npx @modelcontextprotocol/inspector

Python:

  • Verify syntax: python -m py_compile your_server.py

  • Test with MCP Inspector

See language-specific guides for detailed testing approaches and quality checklists.


Phase 4: Create Evaluations

After implementing your MCP server, create comprehensive evaluations to test its effectiveness.

Load ✅ Evaluation Guide for complete evaluation guidelines.

4.1 Understand Evaluation Purpose

Use evaluations to test whether LLMs can effectively use your MCP server to answer realistic, complex questions.

4.2 Create 10 Evaluation Questions

To create effective evaluations, follow the process outlined in the evaluation guide:

  1. Tool Inspection: List available tools and understand their capabilities

  2. Content Exploration: Use READ-ONLY operations to explore available data

  3. Question Generation: Create 10 complex, realistic questions

  4. Answer Verification: Solve each question yourself to verify answers

4.3 Evaluation Requirements

Ensure each question is:

  • Independent: Not dependent on other questions

  • Read-only: Only non-destructive operations required

  • Complex: Requiring multiple tool calls and deep exploration

  • Realistic: Based on real use cases humans would care about

  • Verifiable: Single, clear answer that can be verified by string comparison

  • Stable: Answer won't change over time

4.4 Output Format

Create an XML file with this structure:

<evaluation>
  <qa_pair>
    <question>Find discussions about AI model launches with animal codenames. One model needed a specific safety designation that uses the format ASL-X. What number X was being determined for the model named after a spotted wild cat?</question>
    <answer>3</answer>
  </qa_pair>
<!-- More qa_pairs... -->
</evaluation>

Reference Files

📚 Documentation Library

Load these resources as needed during development:

Core MCP Documentation (Load First)
  • MCP Protocol: Start with sitemap at https://modelcontextprotocol.io/sitemap.xml, then fetch specific pages with .md suffix

  • 📋 MCP Best Practices - Universal MCP guidelines including:

    • Server and tool naming conventions

    • Response format guidelines (JSON vs Markdown)

    • Pagination best practices

    • Transport selection (streamable HTTP vs stdio)

    • Security and error handling standards

Microsoft MCP Documentation (For Azure/Foundry)
  • 🔷 Microsoft MCP Patterns - Microsoft-specific patterns including:

    • Azure MCP Server architecture (48+ Azure services)

    • C#/.NET command implementation patterns

    • Remote MCP with Foundry Agent Service

    • Authentication (Entra ID, OBO flow, Managed Identity)

    • Testing infrastructure with Bicep templates

SDK Documentation (Load During Phase 1/2)
  • Python SDK: Fetch from https://raw.githubusercontent.com/modelcontextprotocol/python-sdk/main/README.md

  • TypeScript SDK: Fetch from https://raw.githubusercontent.com/modelcontextprotocol/typescript-sdk/main/README.md

  • Microsoft MCP SDK: See Microsoft MCP Patterns for C#/.NET

Language-Specific Implementation Guides (Load During Phase 2)
  • 🐍 Python Implementation Guide - Complete Python/FastMCP guide with:

    • Server initialization patterns

    • Pydantic model examples

    • Tool registration with @mcp.tool

    • Complete working examples

    • Quality checklist

  • ⚡ TypeScript Implementation Guide - Complete TypeScript guide with:

    • Project structure

    • Zod schema patterns

    • Tool registration with server.registerTool

    • Complete working examples

    • Quality checklist

  • 🔷 Microsoft MCP Patterns - Complete C#/.NET guide with:

    • Command hierarchy (BaseCommand → GlobalCommand → SubscriptionCommand)

    • Naming conventions ({Resource}{Operation}Command)

    • Option handling with .AsRequired() / .AsOptional()

    • Azure Functions remote MCP deployment

    • Live test patterns with Bicep

Evaluation Guide (Load During Phase 4)
  • ✅ Evaluation Guide - Complete evaluation creation guide with:

    • Question creation guidelines

    • Answer verification strategies

    • XML format specifications

    • Example questions and answers

    • Running an evaluation with the provided scripts

Files

reference/microsoft_mcp_patterns.md

Microsoft MCP Patterns

This document covers patterns from Microsoft's official MCP implementations, including the Azure MCP Server and Foundry MCP services. Use these patterns when building MCP servers that integrate with Azure services or follow Microsoft conventions.


Quick Reference

Microsoft MCP Ecosystem
ServerTypeTransportDescription
Azure MCPLocalstdio48+ Azure services (Storage, KeyVault, Cosmos, SQL, etc.)
Foundry MCPRemoteStreamable HTTPhttps://mcp.ai.azure.com - Models, deployments, evals, agents
Fabric MCPLocalstdioMicrosoft Fabric APIs, OneLake, item definitions
Playwright MCPLocalstdioBrowser automation and testing
GitHub MCPRemoteStreamable HTTPhttps://api.githubcopilot.com/mcp
Azure DevOpsLocalstdioPipelines, repos, boards
SQL ServerLocalstdioSQL queries, schema (aka.ms/MssqlMcp)
Microsoft LearnRemoteStreamable HTTPhttps://learn.microsoft.com/api/mcp
Distribution Formats

Install Azure MCP Server via your preferred method:

FormatPackage/CommandBest For
VS Code Extensionms-azuretools.vscode-azure-mcp-serverVS Code users
NPMnpx @azure/mcp@latestNode.js environments
.NET Tooldotnet tool install -g Azure.Mcp.NET developers
Dockermcr.microsoft.com/azure-sdk/azure-mcpContainers
Command Naming Convention
azmcp <service> <resource> <operation>

Examples:
- azmcp storage blob list
- azmcp keyvault secret get
- azmcp cosmos document create
- azmcp appconfig kv set
C# Command Class Naming
{Resource}{Operation}Command

Examples:
- StorageContainerListCommand
- KeyVaultSecretGetCommand
- CosmosDbDatabaseQueryCommand
Tool Metadata Annotations
AnnotationPurposeExample Use
DestructiveModifies or deletes resourcesDelete operations
ReadOnlyOnly reads data, no side effectsList, get operations
IdempotentSafe to retry, same resultCreate-or-update operations
OpenWorldInteracts with external systemsAPI calls to Azure
SecretHandles sensitive dataKey Vault operations
LocalRequiredRequires local resourcesFile system operations

Server Architecture

Azure MCP Service Coverage (48 Services)
CategoryServices
ComputeApp Service, Functions, AKS, Container Registry, Virtual Desktop
StorageBlob, File Shares, Cosmos DB, Redis, Managed Lustre
DataSQL, PostgreSQL, MySQL, Data Explorer (Kusto), Confidential Ledger
AI/MLAI Foundry, Search, Speech, Application Insights
IntegrationEvent Grid, Event Hubs, Service Bus, SignalR, Communication Services
SecurityKey Vault, RBAC (Authorization), Policy, Advisor
ManagementMonitor, Resource Health, Workbooks, Quota, Azure Migrate
DevOpsLoad Testing, Grafana, Bicep Schema

Full list: Storage, KeyVault, Cosmos, SQL, Postgres, MySQL, AKS, ACR, AppService, Functions, EventGrid, EventHubs, ServiceBus, SignalR, Search, Speech, Monitor, Advisor, Policy, Authorization, and 28+ more.

Server Modes

The Azure MCP Server supports multiple operational modes:

ModeDescriptionUse Case
namespaceDefault mode, tools organized by service namespaceGeneral use
consolidatedAll tools in a single flat namespaceSimple integrations
allExposes all available toolsDevelopment/testing
singleSingle service onlyFocused integrations
Configuration Example (mcp.json)
{
  "mcpServers": {
    "azure": {
      "command": "npx",
      "args": ["-y", "@azure/mcp@latest", "server", "start"],
      "env": {
        "AZURE_TENANT_ID": "${AZURE_TENANT_ID}",
        "AZURE_CLIENT_ID": "${AZURE_CLIENT_ID}"
      }
    }
  }
}
Server Manifest (server.json)
{
  "name": "azure-mcp-server",
  "version": "1.0.0",
  "description": "MCP server for Azure services",
  "capabilities": {
    "tools": true,
    "resources": false,
    "prompts": false
  },
  "transports": ["stdio"],
  "authentication": {
    "type": "azure-identity",
    "scopes": ["https://management.azure.com/.default"]
  }
}

Modular Architecture Pattern (C#/.NET)

Microsoft's Azure MCP Server uses a modular "Area" pattern for organizing tools by service domain.

Command Hierarchy
IBaseCommand
└── BaseCommand
    └── GlobalCommand<TOptions>              # No subscription required
        └── SubscriptionCommand<TOptions>    # Subscription context
            └── Service-specific commands     # Azure service operations
IAreaSetup Interface

Each Azure service implements IAreaSetup to register its tools:

public interface IAreaSetup
{
    string Name { get; }           // e.g., "storage", "keyvault"
    string Title { get; }          // e.g., "Azure Storage", "Azure Key Vault"
    
    void ConfigureServices(IServiceCollection services);
    void RegisterCommands(CommandGroup rootGroup);
}
Implementation Example
public class StorageSetup : IAreaSetup
{
    public string Name => "storage";
    public string Title => "Azure Storage";

    public void ConfigureServices(IServiceCollection services)
    {
        services.AddSingleton<BlobServiceClientFactory>();
        services.AddSingleton<StorageAccountService>();
    }

    public void RegisterCommands(CommandGroup rootGroup)
    {
        var storageGroup = rootGroup.AddGroup("storage", "Azure Storage operations");
        
        // Blob operations
        var blobGroup = storageGroup.AddGroup("blob", "Blob storage operations");
        blobGroup.AddCommand<ListBlobsCommand>("list", "List blobs in a container");
        blobGroup.AddCommand<GetBlobCommand>("get", "Get blob content");
        blobGroup.AddCommand<UploadBlobCommand>("upload", "Upload a blob");
        blobGroup.AddCommand<DeleteBlobCommand>("delete", "Delete a blob");
        
        // Container operations
        var containerGroup = storageGroup.AddGroup("container", "Container operations");
        containerGroup.AddCommand<ListContainersCommand>("list", "List containers");
        containerGroup.AddCommand<CreateContainerCommand>("create", "Create a container");
    }
}
Command Implementation
public class ListBlobsCommand : ICommand
{
    [Option("--account", "Storage account name")]
    public string AccountName { get; set; } = string.Empty;
    
    [Option("--container", "Container name")]
    public string ContainerName { get; set; } = string.Empty;
    
    [Option("--prefix", "Blob name prefix filter")]
    public string? Prefix { get; set; }
    
    [Option("--limit", "Maximum number of results")]
    public int Limit { get; set; } = 100;

    public ToolMetadata Metadata => new()
    {
        ReadOnly = true,
        OpenWorld = true,
        Idempotent = true
    };

    public async Task<CommandResponse> ExecuteAsync(
        IServiceProvider services,
        CancellationToken cancellationToken)
    {
        var factory = services.GetRequiredService<BlobServiceClientFactory>();
        var client = factory.CreateClient(AccountName);
        var container = client.GetBlobContainerClient(ContainerName);
        
        var blobs = new List<BlobItem>();
        await foreach (var blob in container.GetBlobsAsync(prefix: Prefix, cancellationToken: cancellationToken))
        {
            blobs.Add(blob);
            if (blobs.Count >= Limit) break;
        }
        
        return CommandResponse.Success(new
        {
            account = AccountName,
            container = ContainerName,
            count = blobs.Count,
            blobs = blobs.Select(b => new
            {
                name = b.Name,
                size = b.Properties.ContentLength,
                lastModified = b.Properties.LastModified,
                contentType = b.Properties.ContentType
            })
        });
    }
}
Option Registration Extensions
// Required options
command.AddOption("--account")
    .AsRequired()
    .WithDescription("Storage account name");

// Optional options with defaults
command.AddOption("--limit")
    .AsOptional()
    .WithDefault(100)
    .WithDescription("Maximum results to return");

// Enum options
command.AddOption("--access-tier")
    .AsOptional()
    .WithValues<AccessTier>()
    .WithDescription("Blob access tier");

Authentication Patterns

Azure Identity Credential Chain

Microsoft uses a credential chain that tries multiple authentication methods:

1. Environment Variables (AZURE_CLIENT_ID, AZURE_CLIENT_SECRET, AZURE_TENANT_ID)
2. Visual Studio credentials
3. Azure CLI credentials
4. Azure PowerShell credentials
5. Azure Developer CLI credentials
6. Interactive browser (development only)
Environment Variable Controls
VariablePurpose
AZURE_TOKEN_CREDENTIALSOverride credential type
AZURE_TENANT_IDAzure AD tenant ID
AZURE_CLIENT_IDService principal client ID
AZURE_CLIENT_SECRETService principal secret
AZURE_SUBSCRIPTION_IDDefault subscription
Credential Implementation
public class AzureCredentialProvider
{
    public TokenCredential GetCredential()
    {
        var credentialType = Environment.GetEnvironmentVariable("AZURE_TOKEN_CREDENTIALS");
        
        return credentialType?.ToLower() switch
        {
            "environment" => new EnvironmentCredential(),
            "cli" => new AzureCliCredential(),
            "managedidentity" => new ManagedIdentityCredential(),
            _ => new DefaultAzureCredential(new DefaultAzureCredentialOptions
            {
                ExcludeInteractiveBrowserCredential = !IsDevelopment(),
                ExcludeManagedIdentityCredential = IsDevelopment()
            })
        };
    }
}

Error Handling Patterns

Azure SDK Error Handling
public static CommandResponse HandleAzureError(RequestFailedException ex)
{
    return ex.Status switch
    {
        404 => CommandResponse.Error(
            $"Resource not found. Verify the resource exists and you have access. " +
            $"Details: {ex.Message}"),
        
        403 => CommandResponse.Error(
            $"Access denied. Check your permissions and ensure your credentials " +
            $"have the required role assignments. Details: {ex.Message}"),
        
        401 => CommandResponse.Error(
            $"Authentication failed. Run 'az login' or check your credentials. " +
            $"Details: {ex.Message}"),
        
        409 => CommandResponse.Error(
            $"Conflict: Resource already exists or is in a conflicting state. " +
            $"Details: {ex.Message}"),
        
        429 => CommandResponse.Error(
            $"Rate limit exceeded. Wait and retry. " +
            $"Details: {ex.Message}"),
        
        _ => CommandResponse.Error(
            $"Azure API error (HTTP {ex.Status}): {ex.Message}")
    };
}
CommandResponse Pattern
public class CommandResponse
{
    public bool IsSuccess { get; init; }
    public object? Data { get; init; }
    public string? Error { get; init; }
    
    public static CommandResponse Success(object data) => 
        new() { IsSuccess = true, Data = data };
    
    public static CommandResponse Error(string message) => 
        new() { IsSuccess = false, Error = message };
}

Tool Metadata Best Practices

Annotation Guidelines
Operation TypeDestructiveReadOnlyIdempotentOpenWorld
List/Get❌✅✅✅
Create❌❌❌✅
Create-or-Update❌❌✅✅
Update❌❌✅✅
Delete✅❌✅✅
Purge (permanent)✅❌❌✅
Metadata Implementation
public class ToolMetadata
{
    /// <summary>
    /// Tool may perform destructive operations (delete, purge).
    /// Clients should confirm before execution.
    /// </summary>
    public bool Destructive { get; init; }
    
    /// <summary>
    /// Tool only reads data, no side effects.
    /// Safe to call without confirmation.
    /// </summary>
    public bool ReadOnly { get; init; }
    
    /// <summary>
    /// Repeated calls with same arguments produce same result.
    /// Safe to retry on failure.
    /// </summary>
    public bool Idempotent { get; init; }
    
    /// <summary>
    /// Tool interacts with external systems (Azure APIs).
    /// May have latency or availability concerns.
    /// </summary>
    public bool OpenWorld { get; init; }
    
    /// <summary>
    /// Tool handles sensitive data (secrets, keys, credentials).
    /// Output should be treated as confidential.
    /// </summary>
    public bool Secret { get; init; }
    
    /// <summary>
    /// Tool requires local resources (file system, environment).
    /// May not work in all execution contexts.
    /// </summary>
    public bool LocalRequired { get; init; }
}

Remote MCP with Azure Functions

Initialize from Template
# Python
azd init --template remote-mcp-functions-python -e mcpserver-python

# TypeScript  
azd init --template remote-mcp-functions-typescript -e mcpserver-ts

# .NET
azd init --template remote-mcp-functions-dotnet -e mcpserver-dotnet
TypeScript Tool Registration (Azure Functions)
import { app, InvocationContext } from '@azure/functions';

export async function mcpToolStorageList(
    _toolArguments: unknown, 
    context: InvocationContext
): Promise<string> {
    const args = context.triggerMetadata.mcptoolargs as { 
        accountName: string;
        prefix?: string;
    };
    
    // Implementation
    const containers = await listContainers(args.accountName, args.prefix);
    return JSON.stringify(containers);
}

app.mcpTool('storage_container_list', {
    toolName: 'storage_container_list',
    description: 'List all containers in an Azure Storage account',
    toolProperties: {
        accountName: arg.string()
            .describe('Storage account name')
            .required(),
        prefix: arg.string()
            .describe('Filter containers by prefix')
            .optional()
    },
    handler: mcpToolStorageList
});
Deploy to Azure
# Deploy
azd up

# Get endpoint
# https://{function_app_name}.azurewebsites.net/runtime/webhooks/mcp

Testing Infrastructure

Test Structure
tools/Azure.Mcp.Tools.{Service}/tests/
├── Azure.Mcp.Tools.{Service}.UnitTests/    # No Azure resources
├── Azure.Mcp.Tools.{Service}.LiveTests/    # Requires Azure
├── test-resources.bicep                     # Infrastructure template
└── test-resources-post.ps1                  # Post-deployment setup
Unit Tests (No Azure Resources)
[TestClass]
public class StorageContainerListCommandTests
{
    [TestMethod]
    public async Task ListContainers_ReturnsExpectedResults()
    {
        // Arrange
        var mockClient = new Mock<BlobServiceClient>();
        var command = new StorageContainerListCommand(mockClient.Object);
        
        // Act
        var result = await command.ExecuteAsync(new StorageContainerListOptions
        {
            AccountName = "teststorage"
        });
        
        // Assert
        Assert.IsNotNull(result);
    }
}
Live Tests (Bicep Infrastructure)
// test-resources.bicep
param baseName string
param location string = resourceGroup().location

resource storageAccount 'Microsoft.Storage/storageAccounts@2023-01-01' = {
  name: '${baseName}storage'
  location: location
  sku: { name: 'Standard_LRS' }
  kind: 'StorageV2'
}

output STORAGE_ACCOUNT_NAME string = storageAccount.name
Running Tests
# Unit tests only
./eng/scripts/Test-Code.ps1 -UnitTests

# Live tests (deploys Azure resources)
./eng/scripts/Test-Code.ps1 -LiveTests -ServiceDirectory Storage

# Deploy test resources manually
./eng/scripts/Deploy-TestResources.ps1 -ServiceDirectory Storage

Engineering Standards

Code Style Requirements
RuleRequirement
ConstructorsUse C# primary constructors
JSONUse System.Text.Json (no Newtonsoft)
ClassesMake sealed when not inherited
MembersMake static when possible
FilesOne class/interface per file
AOTAll code must be Native AOT compatible
AOT Compatibility
// GOOD: AOT compatible
[JsonSerializable(typeof(ContainerListResponse))]
internal partial class ContainerJsonContext : JsonSerializerContext { }

// BAD: Reflection-based (not AOT safe)
JsonSerializer.Deserialize<ContainerListResponse>(json);  // Avoid
PR Checklist

Before submitting changes to Microsoft MCP:

  • [ ] Code builds with dotnet build

  • [ ] All unit tests pass

  • [ ] Live tests pass (if applicable)

  • [ ] AOT analysis passes: ./eng/scripts/Analyze-AOT-Compact.ps1

  • [ ] Tool descriptions are clear and specific

  • [ ] No hardcoded credentials or endpoints

  • [ ] Follows naming conventions


Integration with Foundry Agent Service

MCP Tool Declaration (Python)
from azure.ai.projects import AIProjectClient
from azure.ai.projects.models import PromptAgentDefinition, MCPTool

mcp_tool = MCPTool(
    server_label="azure-storage",
    server_url="https://{app}.azurewebsites.net/runtime/webhooks/mcp",
    require_approval="always",
    allowed_tools=["storage_container_list"],
    project_connection_id=connection_name
)

agent = project_client.agents.create_version(
    agent_name="storage-agent",
    definition=PromptAgentDefinition(
        model="gpt-4o-mini",
        instructions="Help users manage Azure Storage",
        tools=[mcp_tool]
    )
)
MCP Tool Declaration (C#)
PromptAgentDefinition agentDefinition = new(model: "gpt-4o-mini")
{
    Instructions = "Help users manage Azure Storage",
    Tools = { 
        ResponseTool.CreateMcpTool(
            serverLabel: "azure-storage",
            serverUri: new Uri("https://{app}.azurewebsites.net/runtime/webhooks/mcp"),
            toolCallApprovalPolicy: new McpToolCallApprovalPolicy(
                GlobalMcpToolCallApprovalPolicy.AlwaysRequireApproval
            )
        )
    }
};
Handling Approval Requests
for item in response.output:
    if item.type == "mcp_approval_request" and item.id:
        print(f"Tool: {getattr(item, 'name', '<unknown>')}")
        print(f"Arguments: {json.dumps(getattr(item, 'arguments', None), indent=2)}")
        
        should_approve = input("Approve? (y/N): ").strip().lower() == "y"
        
        input_list.append(McpApprovalResponse(
            type="mcp_approval_response",
            approve=should_approve,
            approval_request_id=item.id
        ))

When to Use Microsoft MCP vs Custom

Use Microsoft MCP Servers When:
  • ✅ Integrating with Azure services (Storage, KeyVault, Cosmos, etc.)

  • ✅ Working with AI Foundry (agents, evaluations, deployments)

  • ✅ Need browser automation (Playwright MCP)

  • ✅ GitHub integration (GitHub MCP)

  • ✅ Microsoft Fabric operations

Build Custom MCP Servers When:
  • ✅ Integrating with non-Microsoft services

  • ✅ Custom internal APIs

  • ✅ Third-party SaaS integrations

  • ✅ Specialized domain logic not covered by existing servers

  • ✅ Need to extend Azure MCP with custom tools

Extending Azure MCP

If you need to add custom tools alongside Azure MCP:

  1. Separate Server: Create a custom MCP server for your tools

  2. Multi-Server Config: Configure both servers in mcp.json

  3. Namespace Prefix: Use distinct prefixes to avoid conflicts

{
  "mcpServers": {
    "azure": {
      "command": "npx",
      "args": ["-y", "@azure/mcp@latest", "server", "start"]
    },
    "custom": {
      "command": "node",
      "args": ["./dist/custom-mcp-server.js"]
    }
  }
}

Resources

Code Samples
SampleDescriptionLanguage
remote-mcp-apim-functions-pythonSecure remote MCP with API ManagementPython
remote-mcp-functions-pythonBasic Azure Functions MCPPython
remote-mcp-functions-dotnetAzure Functions MCPC#
remote-mcp-functions-typescriptAzure Functions MCPTypeScript
mcp-container-tsMCP on Azure Container AppsTypeScript

Requirements and compatibility

Written for: Claude Code

Works with: Claude Code, Claude apps, OpenAI Codex, GitHub Copilot, Cursor, Gemini CLI, Kiro, OpenCode, Goose, Amp, Roo Code, JetBrains Junie, Factory Droid, Devin (ex-Windsurf), Mistral Vibe, OpenHands, Letta

“Works with” means the tool reads the open Agent Skills format; a skill written for one tool may use features others lack.

Needs: Names Claude Code tools, Needs an MCP server, Needs network access, Needs Python, Needs Node.js

Security

What our scanner noticed· Low risk

  • Instructions that use the network
  • Scripts that your agent may run

This skill includes scripts that your agent may run. Read them before you use it.

A static scan is not a guarantee. Skills are third-party content: review a skill before you let an agent use it.

Attribution and licence

Published by
Microsoft
Publisher
Microsoft (Official)
Source
microsoft/skills · .github/skills/mcp-builder · commit 0696600
Licence
MIT License (redistribution allowed)
  • LICENSE (added by AutoGineer from LICENSE)
  • ATTRIBUTION.md (written by AutoGineer)
Synced

AutoGineer is not affiliated with or endorsed by Microsoft.

Download ZIP

The ZIP holds the skill’s files, its licence files and an ATTRIBUTION.md. Download mcp-builder.zip