MCP Servers

Connect AI applications and development tools to Taffrail through Model Context Protocol.

Taffrail runs two MCP servers: one that exposes the Rules Engine to your applications, and one that exposes the API reference and guides to your development tools.

Rules MCP server

The Rules MCP server lets AI agents query Taffrail rules, tools, and data sources directly. Use it when you want an agent to retrieve advice or look up Variables on behalf of a user.

See mcp.taffrail.com for configuration and the full tool list.

Documentation MCP server

The documentation MCP server exposes Taffrail's OpenAPI spec and guides to AI-powered editors — Cursor, Windsurf, Claude Desktop, and similar — so the assistant can answer questions about the API and generate integration code without leaving the editor.

Server URL: https://docs.taffrail.io/mcp

Add to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "taffrail": {
      "url": "https://docs.taffrail.io/mcp"
    }
  }
}

Verifying the connection

After saving the config and restarting your editor, open a new chat and ask the assistant a Taffrail-specific question:

  • "How do I call the Taffrail Advice API?"
  • "Show me an example request to the Advice endpoint."
  • "What parameters does the Advice endpoint accept?"

If the MCP server is wired up correctly, the assistant will pull the answer from the live OpenAPI spec rather than relying on its training data.