modelcontextprotocol

At the start of every response, say "APPLIED MODELCONTEXTPROTOCOL RULE".

Model Context Protocol (MCP) Server Development Rules and Guidelines

Overview

This document provides comprehensive rules, guidelines, and best practices for developing Model Context Protocol (MCP) servers, based on the official MCP specification and the Apache Airflow MCP server implementation.

Core MCP Concepts

1. MCP Server Capabilities

MCP servers can provide three main types of capabilities:

1. Resources: File-like data that can be read by clients (like API responses or file contents)
2. Tools: Functions that can be called by the LLM (with user approval)
3. Prompts: Pre-written templates that help users accomplish specific tasks

2. Architecture Principles

• Stateless Design: Servers should be stateless and handle each request independently
• Standard Transport: Use stdio or SSE transport protocols as defined in the MCP specification
• Type Safety: Leverage Python type hints for automatic tool definition generation
• Async First: Use asynchronous programming patterns for better performance

Development Guidelines

1. Project Structure

Follow this recommended project structure:

project-root/
├── src/
│   ├── __init__.py
│   ├── __main__.py
│   ├── main.py              # CLI entry point
│   ├── server.py            # FastMCP server instance
│   ├── enums.py             # Project enums
│   ├── envs.py              # Environment configuration
│   └── [domain]/            # Domain-specific modules
│       ├── __init__.py
│       ├── client.py        # External API client setup
│       └── [feature].py     # Feature-specific tool implementations
├── pyproject.toml
├── README.md
├── Makefile
└── @modelcontextprotocol.mdc  # This file

2. Tool Implementation Standards

Tool Function Signature

All tool functions must follow this pattern:

from typing import Any, List, Optional, Union
import mcp.types as types

async def tool_name(
    required_param: str,
    optional_param: Optional[str] = None
) -> List[Union[types.TextContent, types.ImageContent, types.EmbeddedResource]]:
    """
    Brief description of what the tool does.
    
    Args:
        required_param: Description of required parameter
        optional_param: Description of optional parameter
    
    Returns:
        List of MCP content types
    """
    # Implementation here
    return [types.TextContent(type="text", text=str(result))]