MCPnet - A Network for Multimodal Capability Protocol (MCP) Servers

MCPnet is a .NET client library that allows AI applications to connect with MCP servers through a centralized network. This library provides a clean, typed interface for discovering and invoking MCP capabilities across the network.

What is MCP?

MCP (Multimodal Capability Protocol) is an open standard for AI applications (called "MCP Clients") to connect with MCP Servers. These servers provide a set of "tools" that AI applications can access, greatly extending their capabilities.

What is MCPnet?

MCPnet is a centralized network of MCP Servers that makes it frictionless to find and use MCP capabilities. The network includes:

• Discovery of MCP servers and capabilities
• Ratings and reviews of servers
• Semantic search for finding the right tools
• Creation of server "remixes" that combine tools from different servers
• And much more...

Think of it as the "Hugging Face of MCP" - a way to discover and connect to MCP Servers.

Installation

You can install the MCPnet package via NuGet:

dotnet add package MCPnet

Usage

Basic Setup

// Add the MCPnet services to your DI container
services.AddMcpServices();

// Or configure them individually
services.AddMcpServices(
    registryBaseUrl: "https://api.mcp.net",
    configureMcpClient: builder => builder.ConfigureHttpClient(client =>
    {
        client.Timeout = TimeSpan.FromSeconds(30);
    }));

Discovering MCP Servers

// Get the registry client
var registry = serviceProvider.GetRequiredService<IMcpRegistry>();

// Search for servers related to image processing
var servers = await registry.SearchServersAsync(
    query: "image processing",
    minRating: 4.0);

foreach (var server in servers)
{
    Console.WriteLine($"{server.Name}: {server.Description}");
}

Invoking MCP Capabilities

// Get the MCP client
var mcpClient = serviceProvider.GetRequiredService<IMcpClient>();

// Discover capabilities on a server
var capabilities = await mcpClient.DiscoverCapabilitiesAsync(serverUrl);

// Invoke a capability
var request = new McpRequest
{
    CapabilityId = "image-generation",
    Input = new Dictionary<string, object>
    {
        { "prompt", "A beautiful sunset over mountains" },
        { "style", "photorealistic" }
    }
};

var response = await mcpClient.InvokeCapabilityAsync(serverUrl, request);

if (response.Success)
{
    var imageUrl = response.Output["image_url"] as string;
    Console.WriteLine($"Generated image: {imageUrl}");
}

Creating Capability Remixes

// Create a remix of capabilities from different servers
var remix = await registry.CreateRemixAsync(
    name: "Advanced Text Processing Suite",
    description: "A collection of the best text processing tools",
    capabilities: new List<(string ServerId, string CapabilityId)>
    {
        ("server1", "text-summarization"),
        ("server2", "sentiment-analysis"),
        ("server3", "language-translation")
    });

Console.WriteLine($"Created remix: {remix.Name} with {remix.Capabilities.Count} capabilities");

Industry-Specific Examples

MCPnet can be applied across various industries to solve domain-specific problems:

Healthcare

// Find medical imaging analysis capabilities
var medicalServers = await mcpRegistry.SearchServersAsync(
    query: "medical imaging analysis radiology",
    minRating: 4.5);

// Example: Process a CT scan for lung nodule detection
var request = new McpRequest
{
    CapabilityId = lungNoduleCapability.Id,
    Input = new Dictionary<string, object>
    {
        { "image_data", ctScanData },
        { "patient_age", 65 },
        { "patient_sex", "male" },
        { "scan_type", "chest_ct" }
    }
};

var response = await mcpClient.InvokeCapabilityAsync(medicalServer.Url, request);

Finance

// Create a custom financial analysis toolkit by remixing capabilities
var capabilitiesToRemix = new List<(string ServerId, string CapabilityId)>();

// Add sentiment analysis for financial news
capabilitiesToRemix.Add((sentimentCapabilities[0].Server.Id, sentimentCapabilities[0].Capability.Id));

// Add stock prediction capabilities
capabilitiesToRemix.Add((predictionCapabilities[0].Server.Id, predictionCapabilities[0].Capability.Id));

// Add fraud detection capabilities
capabilitiesToRemix.Add((fraudCapabilities[0].Server.Id, fraudCapabilities[0].Capability.Id));

var financeToolkit = await mcpRegistry.CreateRemixAsync(
    name: "Advanced Financial Analysis Toolkit",
    description: "A comprehensive toolkit for financial analysis, prediction and fraud detection",
    capabilities: capabilitiesToRemix);

Content Creation

// Generate images based on a product description
var imageRequest = new McpRequest
{
    CapabilityId = imageCapability.Id,
    Input = new Dictionary<string, object>
    {
        { "prompt", "A sleek modern smartphone with an edge-to-edge display" },
        { "style", "photorealistic product photography" },
        { "aspect_ratio", "16:9" },
        { "count", 3 }
    }
};

// Generate narration audio for promotional content
var audioRequest = new McpRequest
{
    CapabilityId = audioCapability.Id,
    Input = new Dictionary<string, object>
    {
        { "text", "Introducing our latest smartphone..." },
        { "voice", "professional_male_en_us" },
        { "speed", 1.0 }
    }
};

// Combine images and audio into a video
var videoRequest = new McpRequest
{
    CapabilityId = videoCapability.Id,
    Input = new Dictionary<string, object>
    {
        { "images", imageUrls },
        { "audio_track", audioUrl },
        { "duration", 30 }
    }
};

Enterprise Integration Examples

MCPnet integrates seamlessly with enterprise systems:

Document Processing Workflow

// Process an invoice through multiple stages
// Step 1: Convert document to text using OCR
var ocrRequest = new McpRequest
{
    CapabilityId = ocrCapability.Id,
    Input = new Dictionary<string, object>
    {
        { "document_url", invoiceDocument },
        { "output_format", "text" },
        { "language", "en" }
    }
};

// Step 2: Classify the document type
var classificationRequest = new McpRequest
{
    CapabilityId = classificationCapability.Id,
    Input = new Dictionary<string, object>
    {
        { "text", textContent },
        { "possible_types", new[] { "invoice", "purchase_order", "receipt" } }
    }
};

// Step 3: Extract structured data based on document type
var extractionRequest = new McpRequest
{
    CapabilityId = extractionCapability.Id,
    Input = new Dictionary<string, object>
    {
        { "text", textContent },
        { "document_type", documentType },
        { "fields", new[] { "invoice_number", "date", "total_amount" } }
    }
};

IoT Data Analysis

// Detect anomalies in sensor data
var anomalyRequest = new McpRequest
{
    CapabilityId = anomalyCapability.Id,
    Input = new Dictionary<string, object>
    {
        { "sensor_data", sensorData },
        { "sensitivity", "medium" }
    }
};

// Predict equipment maintenance needs
var maintenanceRequest = new McpRequest
{
    CapabilityId = maintenanceCapability.Id,
    Input = new Dictionary<string, object>
    {
        { "equipment_id", "PUMP-123" },
        { "sensor_data", sensorData },
        { "anomalies", anomalies }
    }
};

SAP Integration for Supply Chain Management

// Generate demand forecasts
var forecastRequest = new McpRequest
{
    CapabilityId = forecastCapability.Id,
    Input = new Dictionary<string, object>
    {
        { "historical_data", historicalSalesData },
        { "forecast_periods", 6 } // 6 months ahead
    }
};

// Optimize inventory based on forecasts
var inventoryRequest = new McpRequest
{
    CapabilityId = inventoryCapability.Id,
    Input = new Dictionary<string, object>
    {
        { "forecasts", forecasts },
        { "current_inventory", currentInventory },
        { "service_level", 0.95 } // 95% service level
    }
};

// Analyze supply chain risks
var riskRequest = new McpRequest
{
    CapabilityId = riskCapability.Id,
    Input = new Dictionary<string, object>
    {
        { "supply_chain_data", supplyChainData },
        { "external_risk_factors", externalRiskFactors }
    }
};

Running the Examples

The library includes a comprehensive set of examples demonstrating how MCPnet can be used in various scenarios. To run them, use the ExampleRunner class:

// Run all examples
await ExampleRunner.RunAllExamplesAsync();

// Run only industry-specific examples
await ExampleRunner.RunIndustryExamplesAsync();

// Run only enterprise integration examples
await ExampleRunner.RunEnterpriseExamplesAsync();

// Run a specific example
await IndustrySpecificExamples.HealthcareExampleAsync();
await EnterpriseIntegrationExamples.IoTDataAnalysisAsync();

Repository

The source code for MCPnet is available on GitHub: https://github.com/adhammehdawi/MCPnet

License

MIT

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.