MCP Servers

The BoxLang AI Module provides a complete MCP (Model Context Protocol) server implementation that allows you to expose tools, resources, and prompts to AI clients.

What is an MCP Server?

An MCP Server is a service that exposes capabilities to AI clients using the standardized Model Context Protocol. It enables:

• 🔧 Expose Tools: Register functions that AI clients can invoke

• 📚 Serve Resources: Provide documents and data to AI clients

• 💬 Offer Prompts: Define reusable prompt templates

• 🌐 HTTP & STDIO Transports: Expose your MCP server via web or command-line

🏗️ MCP Architecture

Documentation

Topic	Description
🚀 Getting Started	Build your first server in 5 minutes
🌐 Transports	HTTP vs STDIO — choose what fits your deployment
🔧 Server Configuration	Authentication, CORS, IP allow lists, body limits, API keys
🧩 Registering Tools, Resources & Prompts	Manual registration with inline MCPServer calls
✨ Annotation-Based Discovery	Auto-register with @mcpTool, @mcpResource, @mcpPrompt
🏗️ Class-Based Servers	Extend MCPServer for better organization
🌍 HTTP Endpoint & Routing	Set up custom HTTP endpoints
📊 Observability & Monitoring	Statistics, events, and real-time monitoring
⏸️ Pause & Resume	Temporarily halt servers for maintenance
⭐ Best Practices	Security, design patterns, production setup
💡 Examples & Use Cases	Complete working code examples

Quick Start

// Application.bx
class {

    function onApplicationStart() {
        MCPServer( "myApp" )
            .setDescription( "My Application MCP Server" )
            .setVersion( "1.0.0" )
            .registerTool(
                aiTool( "search", "Search documents", ( query ) => {
                    return searchService.search( query )
                } )
            )
    }

}

Access at: POST http://localhost/~bxai/mcp.bxm?server=myApp

👉 Full Getting Started Guide →

External Resources

• Model Context Protocol Specification

• MCP Implementation Examples