MCP Protocol
Model Context Protocol: an open standard enabling AI models to securely communicate with external tools.
The Model Context Protocol (MCP) is an open standard developed by Anthropic that enables AI models to communicate with external data sources and tools in a standardized way. It's comparable to a USB-C port for AI: one universal connector for various services.
How does MCP work?
MCP defines a client-server architecture. The AI application (client) connects with MCP servers that provide access to specific data or functionality. These can be databases, APIs, file systems, or web services. Communication is standardized and secure.
MCP and websites
Websites can offer an MCP server so AI agents can directly use their content and services. This is a new layer of AI readiness on top of traditional SEO — bibliotheekterm. The scanner checks whether a website offers MCP discovery signals.
Example: MCP Server Manifest
An MCP server makes itself known via a manifest (JSON) that describes which tools and resources are available. Below is a simplified example:
{
"name": "example-company-mcp",
"version": "1.0.0",
"description": "MCP server for Example Company products and documentation",
"tools": [
{
"name": "search_products",
"description": "Search products by name, category, or features",
"inputSchema": {
"type": "object",
"properties": {
"query": {
"type": "string",
"description": "Search term for products"
},
"category": {
"type": "string",
"description": "Product category filter"
}
},
"required": ["query"]
}
},
{
"name": "get_product_details",
"description": "Get detailed information about a specific product",
"inputSchema": {
"type": "object",
"properties": {
"product_id": {
"type": "string",
"description": "Unique product ID"
}
},
"required": ["product_id"]
}
},
{
"name": "check_availability",
"description": "Check product availability and delivery time",
"inputSchema": {
"type": "object",
"properties": {
"product_id": {
"type": "string",
"description": "Unique product ID"
},
"postal_code": {
"type": "string",
"description": "Postal code for delivery time calculation"
}
},
"required": ["product_id"]
}
}
],
"resources": [
{
"uri": "docs://api-reference",
"name": "API Reference",
"description": "Complete API documentation",
"mimeType": "text/markdown"
}
]
}MCP Discovery for websites
Websites make their MCP server discoverable via one of the following methods:
- Well-known endpoint: a
/.well-known/mcp.jsonfile describing the MCP server URL and capabilities - DNS TXT record: a DNS record pointing to the MCP server
- HTML meta tag: a
<link rel="mcp-server">tag in the page head - HTTP header: an
MCP-Serverresponse header with the server URL
What does our scanner check?
The scanner checks whether your website offers MCP discovery signals: a /.well-known/mcp.json endpoint, relevant HTTP headers, meta tags, or DNS records. MCP discovery is part of the Agent Readiness score.
Frequently asked questions
Does every website need an MCP server?
No, not every website benefits from an MCP server. It's especially relevant for websites with dynamic data, services, or APIs that AI agents would want to use directly. Think of e-commerce (product search, availability), SaaS platforms (account information, performing actions), and knowledge bases (searching, filtering).
Is MCP secure? Who has access?
MCP is designed with security as a core principle. The server determines which tools and data are available and can require authentication via OAuth. You have full control over what is exposed. Sensitive operations can be placed behind authentication.
How does MCP relate to a regular REST API?
A REST API is designed for developers who know and call specific endpoints. MCP is designed for AI agents that need to discover what's available and how to use it themselves. MCP provides self-describing tools with input/output schemas, so an AI agent can determine which tool to use without human help.
Which AI platforms support MCP?
MCP is supported by Claude (Anthropic), and the ecosystem is growing rapidly. More and more AI tools and platforms are implementing MCP client support. The protocol is open source, which encourages broad adoption.
Can I implement MCP in my existing Laravel/WordPress site?
Yes. For Laravel, packages are available that add MCP server functionality. For WordPress, you can build a custom plugin or set up an MCP proxy. Implementation requires defining tools (with input/output schemas) and setting up a transport layer (HTTP or WebSocket).
