MCP Server

What is MCP?

The Model Context Protocol (MCP) is a standard way for an AI application to discover and use external tools.

Our public MCP server lets an AI agent look up the documentation and generated API reference while it helps you build an integration.

The server is read-only. It does not expose your API token, modify Onderwijsregio API data, or make API requests on your behalf.

What can an agent do?

The server is available at https://docs.onderwijsregio.onderwijsin.nl/mcp over HTTP transport.

Use the discovery tools first, then retrieve the specific reference entry you need:

• list-api-tags lists the functional groups in the API reference.
• get-api-tag retrieves one tag and its associated operation summaries.
• list-api-operations lists documented HTTP operations.
• get-api-operation retrieves one operation by method and path, including request and response details.
• list-api-models lists the named API schemas.
• get-api-model retrieves one schema and its fields.

Each reference result includes a link to the corresponding page on this documentation site. This makes the MCP server useful for questions such as:

• Which endpoint should I use to retrieve articles or programmes?
• What fields and relationships does a model contain?
• Which operations are grouped under authentication or search?
• Where can I find the complete reference page for an endpoint?

Connect an AI tool

Most MCP-compatible tools ask for the server URL. Use https://docs.onderwijsregio.onderwijsin.nl/mcp and choose HTTP or streamable HTTP when the client offers a transport choice.

Claude Code

Register the remote server from the command line:

claude mcp add --transport http onderwijsregio-api https://docs.onderwijsregio.onderwijsin.nl/mcp

Cursor

Add the server to .cursor/mcp.json in your project:

{
  "mcpServers": {
    "onderwijsregio-api": {
      "type": "http",
      "url": "https://docs.onderwijsregio.onderwijsin.nl/mcp"
    }
  }
}

Visual Studio Code and other clients

Add an HTTP MCP server to the client configuration using this endpoint:

{
  "name": "onderwijsregio-api",
  "type": "http",
  "url": "https://docs.onderwijsregio.onderwijsin.nl/mcp"
}

Configuration names and file locations differ between clients. Follow the client’s MCP setup instructions if it uses a different configuration shape.

Use the server effectively

Ask the agent to consult the MCP server before it invents an endpoint, field name, or response shape.

For a broad question, ask it to list tags, operations, or models first. For a known endpoint or schema, ask it to retrieve the exact reference entry directly.

The MCP server provides documentation context; it does not replace authentication. Your application still needs an API account, token, and the permissions required for the data you request.

Related resources

• LLMs.txt provides documentation files that AI tools can ingest as context.
• Request API access explains how to obtain an API key and authenticate requests.
• API reference contains the generated endpoint and schema pages.