MCP Server

Ontology Workbench exposes a Model Context Protocol (MCP) server at /api/mcp. This lets AI coding assistants — Claude, Cursor, Windsurf, and any MCP-compatible client — query your documentation and schema data directly.

No authentication is required. The MCP server is publicly accessible.

Endpoint

POST /api/mcp
GET  /api/mcp

• GET /api/mcp — returns discovery metadata (server name, version, available tools)
• POST /api/mcp — executes MCP JSON-RPC 2.0 requests

Available tools

The server exposes three tools:

search_docs

Full-text search across all Ontology Workbench documentation pages.

Input:

{
  "query": "string"
}

Output: An array of matching excerpts with page titles and slugs.

Example:

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "search_docs",
    "arguments": { "query": "export prisma schema" }
  }
}

get_entities

Returns the list of entity type names from a published OWB model.

Input:

{
  "modelId": "string (UUID)"
}

Output: Array of entity type names in the model.

get_relationship

Returns relationship details for a published OWB model.

Input:

{
  "modelId": "string (UUID)"
}

Output: Array of relationships with source, target, and cardinality.

Connecting from Claude Desktop

Add the following to your Claude Desktop MCP configuration (claude_desktop_config.json):

{
  "mcpServers": {
    "ontology-workbench": {
      "url": "https://your-deployment-url.com/api/mcp"
    }
  }
}

Once connected, Claude can search OWB documentation and query your published models during a conversation.

Connecting from Cursor

In Cursor's settings, add a new MCP server with the URL:

https://your-deployment-url.com/api/mcp

Cursor will automatically discover the available tools via the GET /api/mcp discovery endpoint.

Protocol details

The server implements MCP over HTTP using JSON-RPC 2.0. Supported methods:

Discovery response (GET /api/mcp):

{
  "name": "owb-docs",
  "version": "1.0.0",
  "tools": ["search_docs", "get_entities", "get_relationship"]
}

Use cases

• Ask your AI assistant about OWB — "How do I set up API key authentication?" routes to search_docs
• Generate code from your model — give Claude your model ID and ask it to generate a migration, resolver, or seeder using the actual entity types from get_entities
• Cross-reference your schema — ask Cursor to check that a new relationship is consistent with the existing model structure via get_relationship