Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/modelcontextprotocol/csharp-sdk/llms.txt

Use this file to discover all available pages before exploring further.

The MCP C# SDK integrates seamlessly with Microsoft.Extensions.AI, allowing MCP tools to work directly with any IChatClient implementation.

​
Quick Start

Connect to an MCP server and use its tools with an LLM: 
using Microsoft.Extensions.AI;
using ModelContextProtocol.Client;
using OpenAI;

// Connect to MCP server
var transport = new StdioClientTransport(new()
{
    Command = "npx",
    Arguments = ["-y", "@modelcontextprotocol/server-everything"]
});
await using var mcpClient = await McpClient.CreateAsync(transport);

// Get tools from server
var tools = await mcpClient.ListToolsAsync();

// Create LLM client
using IChatClient chatClient = new OpenAIClient(apiKey)
    .AsChatClient("gpt-4o")
    .AsBuilder()
    .UseFunctionInvocation()
    .Build();

// Chat with tools
List<ChatMessage> messages = [
    new(ChatRole.User, "What's the weather in Seattle?")
];

var response = await chatClient.GetResponseAsync(
    messages,
    new ChatOptions { Tools = [.. tools] }
);

Console.WriteLine(response.Message);

​
MCP Tools as AIFunction

MCP tools implement AIFunction from Microsoft.Extensions.AI, making them compatible with any IChatClient: 
await using var client = await McpClient.CreateAsync(transport);

// List all available tools
var tools = await client.ListToolsAsync();
foreach (var tool in tools)
{
    Console.WriteLine($"Tool: {tool.Name}");
    Console.WriteLine($"Description: {tool.Description}");
}

// Tools can be passed directly to ChatOptions
var options = new ChatOptions
{
    Tools = [.. tools]
};

​
Streaming Conversations

Use streaming for real-time responses: 
using Anthropic;
using Microsoft.Extensions.AI;
using ModelContextProtocol.Client;

// Connect to server
var transport = new StdioClientTransport(new()
{
    Command = "dotnet",
    Arguments = ["run", "--project", "../WeatherServer"]
});
await using var mcpClient = await McpClient.CreateAsync(transport);
var tools = await mcpClient.ListToolsAsync();

// Create Anthropic client with function invocation
using var chatClient = new AnthropicClient(apiKey)
    .AsChatClient("claude-4.5-sonnet-20250514")
    .AsBuilder()
    .UseFunctionInvocation()
    .Build();

var messages = new List<ChatMessage>();
var sb = new StringBuilder();

while (true)
{
    Console.Write("You: ");
    var input = Console.ReadLine();
    if (string.IsNullOrWhiteSpace(input)) break;

    messages.Add(new ChatMessage(ChatRole.User, input));

    // Stream response
    await foreach (var update in chatClient.GetStreamingResponseAsync(
        messages,
        new ChatOptions { Tools = [.. tools] }
    ))
    {
        Console.Write(update);
        sb.Append(update.ToString());
    }

    Console.WriteLine();
    messages.Add(new ChatMessage(ChatRole.Assistant, sb.ToString()));
    sb.Clear();
}

​
Multi-Provider Support

The same MCP tools work with any Microsoft.Extensions.AI provider: 
using OpenAI;

var chatClient = new OpenAIClient(apiKey)
    .AsChatClient("gpt-4o")
    .AsBuilder()
    .UseFunctionInvocation()
    .Build();

var response = await chatClient.GetResponseAsync(
    messages,
    new ChatOptions { Tools = [.. tools] }
);

​
Tool Execution

The UseFunctionInvocation() middleware automatically handles tool calls. You can also invoke tools manually: 
await using var client = await McpClient.CreateAsync(transport);

// Call a tool directly
var result = await client.CallToolAsync(
    "get_weather",
    new Dictionary<string, object?>
    {
        ["location"] = "Seattle, WA",
        ["unit"] = "fahrenheit"
    }
);

// Extract text content
var textContent = result.Content
    .OfType<TextContentBlock>()
    .FirstOrDefault()?.Text;

Console.WriteLine(textContent);

​
Progress Reporting

Monitor tool execution progress: 
var progress = new Progress<ProgressNotificationValue>(p =>
{
    Console.WriteLine($"Progress: {p.Progress}/{p.Total}");
});

var result = await client.CallToolAsync(
    "long_running_task",
    arguments: null,
    progress: progress
);

​
Tool Discovery

React to tool list changes: 
await using var disposable = client.RegisterNotificationHandler(
    NotificationMethods.ToolListChangedNotification,
    async (notification, cancellationToken) =>
    {
        Console.WriteLine("Tool list changed, refreshing...");
        var tools = await client.ListToolsAsync();
        // Update your ChatOptions with new tools
    }
);

​
Complete Example

Here’s a complete interactive chat application: 
using Microsoft.Extensions.AI;
using Microsoft.Extensions.Configuration;
using Microsoft.Extensions.Hosting;
using ModelContextProtocol.Client;
using OpenAI;

var builder = Host.CreateApplicationBuilder(args);

builder.Configuration
    .AddEnvironmentVariables()
    .AddUserSecrets<Program>();

// Connect to MCP server
var transport = new StdioClientTransport(new()
{
    Command = "npx",
    Arguments = ["-y", "@modelcontextprotocol/server-everything"],
    Name = "Everything Server"
});
await using var mcpClient = await McpClient.CreateAsync(transport);

// Get available tools
var tools = await mcpClient.ListToolsAsync();
Console.WriteLine($"Connected with {tools.Count} tools available");

// Create LLM client
using var chatClient = new OpenAIClient(
    builder.Configuration["OPENAI_API_KEY"]
)
.AsChatClient("gpt-4o")
.AsBuilder()
.UseFunctionInvocation()
.Build();

var messages = new List<ChatMessage>();
var options = new ChatOptions
{
    Tools = [.. tools]
};

Console.WriteLine("Chat started! Type 'exit' to quit.\n");

while (true)
{
    Console.Write("You: ");
    var input = Console.ReadLine();
    
    if (string.IsNullOrWhiteSpace(input) || 
        input.Equals("exit", StringComparison.OrdinalIgnoreCase))
    {
        break;
    }

    messages.Add(new ChatMessage(ChatRole.User, input));

    var response = await chatClient.GetResponseAsync(messages, options);
    
    Console.WriteLine($"Assistant: {response.Message}");
    messages.Add(response.Message);
}

​
Best Practices

Always enable function invocation: Use .UseFunctionInvocation() in your IChatClient builder to automatically handle tool calls.
Dispose clients properly: Always use await using with McpClient to ensure proper cleanup of server processes and connections.
Tool updates: Register handlers for ToolListChangedNotification to stay synchronized when servers add or remove tools dynamically.

​
Next Steps

Sampling

Handle server sampling requests

Roots

Provide filesystem roots to servers

Build docs developers (and LLMs) love

Get started for freeTalk to us