Skip to main content

Custom MCP Server

For organizations with specific needs, it's possible to set up your own MCP (Model Context Protocol) server. This gives you full control over which functions and data are available to the AI.

What is MCP?

The Model Context Protocol (MCP) is an open standard developed by Anthropic for connecting AI models with external systems. An MCP server exposes "tools" that the AI can use.

Capabilities

With a custom MCP server you can:

  • 🔧 Build custom tools - Add organization-specific functions
  • 🗄️ Connect internal systems - Access databases and APIs
  • 🔒 Control access - Determine precisely what's available
  • 📊 Integrate data - Connect to ERP, CRM, and other systems

Technical Requirements

Server Requirements

  • HTTP(S) endpoint that implements the MCP protocol
  • Authentication (API key, OAuth, or other)
  • JSON-RPC compatible interface

Supported Transports

TransportDescription
HTTP/HTTPSStandard web protocol
WebSocketFor real-time communication
stdioFor local development

Setting Up

Step 1: Develop the MCP Server

Create an MCP server following the official MCP specification:

// Example MCP tool definition
{
"name": "get_inventory",
"description": "Get the current inventory of a product",
"inputSchema": {
"type": "object",
"properties": {
"product_id": {
"type": "string",
"description": "The product ID"
}
},
"required": ["product_id"]
}
}

Step 2: Host the Server

Host your MCP server on:

  • Your own infrastructure
  • Cloud platforms (AWS, Azure, GCP)
  • Serverless platforms (Vercel, Cloudflare Workers)

Step 3: Configure in AI School

  1. Go to Admin Panel > Integrations > MCP Servers
  2. Click Add MCP Server
  3. Enter the server URL
  4. Configure authentication
  5. Test the connection

Example Use Cases

ERP Integration

Build an MCP server that provides access to inventory data, order information, or financial data.

Custom Knowledge Base

Expose internal documentation and FAQ systems as tools.

Business Logic

Add organization-specific calculations, validations, or workflows.

IoT and Sensors

Connect to physical systems and sensors for real-time data.

Security Best Practices

  • ✅ Always use HTTPS
  • ✅ Implement authentication
  • ✅ Validate all inputs
  • ✅ Log all requests
  • ✅ Limit rate and scope
  • ✅ Use principle of least privilege

Resources

WhatsApp