Xeku.MCP 0.0.0.7

Xeku.MCP (Model Context Protocol for XAF)

Xeku.MCP is a bridge module for DevExpress XAF (eXpressApp Framework) that implements the Model Context Protocol (MCP). It allows AI Agents to discover and interact with your XAF Business Objects and custom actions through a standardized JSON-RPC 2.0 interface.

🚀 Key Features

• Standard MCP Protocol: Support for resources/list, tools/list, and tools/call.
• OData Native Proxy: Automatically delegates Create and Update operations to XAF's native OData engine for maximum compatibility, security, and complex business logic execution.
• Recursive Mapping: Supports deep updates and master-detail relationships through nested JSON structures.
• XAF Tracing Integration: Logs all bridge activity directly into the native XAF eXpressAppFramework.log.
• Detailed Error Handling: Captures and returns granular ValidationException results to AI Agents for self-correction.

🏗️ Architecture

graph TD
    subgraph "Presentation (Blue)"
        B1["MCP Controller (API Port)"]:::blue
    end

    subgraph "Service/Logic (Yellow)"
        S1["McpDiscoveryService"]:::yellow
        S2["OData Proxy Engine"]:::yellow
        S3["JSON-RPC Dispatcher"]:::yellow
    end

    subgraph "Data/Entity (Green)"
        D1["XAF Business Objects"]:::green
        D2["ITypesInfo / DC"]:::green
    end

    subgraph "External/Native (Purple)"
        E1["Native XAF OData API"]:::purple
        E2["AI Agent (MCP Client)"]:::purple
    end

    E2 -->|JSON-RPC| B1
    B1 --> S3
    S3 --> S1
    S3 --> S2
    S1 --> D2
    S2 -->|Internal HTTP| E1
    E1 --> D1

    classDef blue fill:#e1f5ff,stroke:#01579b
    classDef yellow fill:#fff4e1,stroke:#f57f17
    classDef green fill:#e8f5e9,stroke:#1b5e20
    classDef purple fill:#f3e5f5,stroke:#4a148c

🤖 AI Agent Integration

Use Cases

• Dynamic Data Entry: Agents can create complex Master-Detail records (e.g., An Order with multiple Items) in a single tool call.
• Automated Workflows: Trigger custom XAF Actions defined on Business Objects.
• Knowledge Discovery: Agents can scan BO metadata to understand available properties and capabilities.

API Integration

AI Agents interact with the bridge via:

• POST /api/mcp/protocol: The main JSON-RPC endpoint for MCP clients.
• GET /api/mcp/resources: REST endpoint for resource discovery.
• GET /api/mcp/tools: REST endpoint for tool discovery.

Best Practices

• Stability: Use the OData Proxy for complex data writes to ensure transaction integrity.
• Error Handling: Monitor the ValidationFailed responses; Agents should be prompted to correct the specific fields mentioned in the Details array.

🛠️ Configuration

Register the MCP services in your Startup.cs:

public void ConfigureServices(IServiceCollection services) {
    // ...
    services.AddXekuMcp();
    services.AddHttpClient(); // Required for OData Proxy
}

📄 License

This module is part of the Xeku private framework.