mcp-advanced-patterns

# MCP Advanced Patterns

Advanced Model Context Protocol patterns for production-grade MCP implementations.

> **FastMCP 2.14.x** (Jan 2026): Enterprise auth, OpenAPI/FastAPI generation, server composition, proxying. Python 3.10-3.13.

## Overview

- Composing multiple tools into orchestrated workflows
- Managing resource lifecycle and caching efficiently
- Scaling MCP servers horizontally with load balancing
- Building custom MCP servers with middleware and transports
- Implementing auto-enable thresholds for context management

## Tool Composition Pattern

```python
from dataclasses import dataclass
from typing import Any, Callable, Awaitable

@dataclass
class ComposedTool:
    """Combine multiple tools into a single pipeline operation."""
    name: str
    tools: dict[str, Callable[..., Awaitable[Any]]]
    pipeline: list[str]

    async def execute(self, input_data: dict[str, Any]) -> dict[str, Any]:
        """Execute tool pipeline sequentially."""
        result = input_data
        for tool_name in self.pipeline:
            tool = self.tools[tool_name]
            result = await tool(result)
        return result

# Example: Search + Summarize composition
search_summarize = ComposedTool(
    name="search_and_summarize",
    tools={
        "search": search_documents,
        "summarize": summarize_content,
    },
    pipeline=["search", "summarize"]
)
```

## FastMCP Server with Lifecycle

```python
from contextlib import asynccontextmanager
from collections.abc import AsyncIterator
from dataclasses import dataclass
from mcp.server.fastmcp import Context, FastMCP

@dataclass
class AppContext:
    """Typed application context with shared resources."""
    db: Database
    cache: CacheService
    config: dict

@asynccontextmanager
async def app_lifespan(server: FastMCP) -> AsyncIterator[AppContext]:
    """Manage server startup and shutdown lifecycle."""
    # Initialize on startup
    db = await Database.connect()
    cache = await CacheService.connect()