This module provides a V language implementation of the Model Context Protocol (MCP) specification. MCP is a protocol designed to standardize communication between AI models and their context providers.

Overview

The MCP module serves as a core library for building MCP-compliant servers in V. Its main purpose is to provide all the boilerplate MCP functionality, so implementers only need to define and register their specific handlers. The module handles the Standard Input/Output (stdio) transport as described in the MCP transport specification, enabling standardized communication between AI models and their context providers.

The module implements all the required MCP protocol methods (resources/list, tools/list, prompts/list, etc.) and manages the underlying JSON-RPC communication, allowing developers to focus solely on implementing their specific tools and handlers. The module itself is not a standalone server but rather a framework that can be used to build different MCP server implementations. The subdirectories within this module (such as baobab and developer) contain specific implementations of MCP servers that utilize this core framework.

to test

curl -fsSL https://bun.sh/install | bash
  source /root/.bashrc

Key Components

Server: The main MCP server struct that handles JSON-RPC requests and responses
Backend Interface: Abstraction for different backend implementations (memory-based by default)
Model Configuration: Structures representing client and server capabilities according to the MCP specification
Protocol Handlers: Implementation of MCP protocol handlers for resources, prompts, tools, and initialization
Factory: Functions to create and configure an MCP server with custom backends and handlers

Features

Complete implementation of the MCP protocol version 2024-11-05
Handles all boilerplate protocol methods (resources/list, tools/list, prompts/list, etc.)
JSON-RPC based communication layer with automatic request/response handling
Support for client-server capability negotiation
Pluggable backend system for different storage and processing needs
Generic type conversion utilities for MCP tool content
Comprehensive error handling
Logging capabilities
Minimal implementation requirements for server developers

Usage

To create a new MCP server using the core module:

import incubaid.herolib.ai.mcp
import incubaid.herolib.schemas.jsonrpc

// Create a backend (memory-based or custom implementation)
backend := mcp.MemoryBackend{
    tools: {
        'my_tool': my_tool_definition
    }
    tool_handlers: {
        'my_tool': my_tool_handler
    }
}

// Create and configure the server
mut server := mcp.new_server(backend, mcp.ServerParams{
    config: mcp.ServerConfiguration{
        server_info: mcp.ServerInfo{
            name: 'my_mcp_server'
            version: '1.0.0'
        }
    }
})!

// Start the server
server.start()!