Merge pull request #203 from modelcontextprotocol/davidsp/clean-lifespan

feat: add lifespan context manager support

Author: dsp-ant

README.md

@@ -128,13 +128,41 @@ The [Model Context Protocol (MCP)](https://modelcontextprotocol.io) lets you bui
 The FastMCP server is your core interface to the MCP protocol. It handles connection management, protocol compliance, and message routing:
 
 ```python
+# Add lifespan support for startup/shutdown with strong typing
+from dataclasses import dataclass
+from typing import AsyncIterator
 from mcp.server.fastmcp import FastMCP
 
 # Create a named server
 mcp = FastMCP("My App")
 
 # Specify dependencies for deployment and development
 mcp = FastMCP("My App", dependencies=["pandas", "numpy"])
+
+@dataclass
+class AppContext:
+    db: Database  # Replace with your actual DB type
+
+@asynccontextmanager
+async def app_lifespan(server: FastMCP) -> AsyncIterator[AppContext]:
+    """Manage application lifecycle with type-safe context"""
+    try:
+        # Initialize on startup
+        await db.connect()
+        yield AppContext(db=db)
+    finally:
+        # Cleanup on shutdown
+        await db.disconnect()
+
+# Pass lifespan to server
+mcp = FastMCP("My App", lifespan=app_lifespan)
+
+# Access type-safe lifespan context in tools
+@mcp.tool()
+def query_db(ctx: Context) -> str:
+    """Tool that uses initialized resources"""
+    db = ctx.request_context.lifespan_context["db"]
+    return db.query()
 ```
 
 ### Resources
@@ -334,7 +362,38 @@ def query_data(sql: str) -> str:
 
 ### Low-Level Server
 
-For more control, you can use the low-level server implementation directly. This gives you full access to the protocol and allows you to customize every aspect of your server:
+For more control, you can use the low-level server implementation directly. This gives you full access to the protocol and allows you to customize every aspect of your server, including lifecycle management through the lifespan API:
+
+```python
+from contextlib import asynccontextmanager
+from typing import AsyncIterator
+
+@asynccontextmanager
+async def server_lifespan(server: Server) -> AsyncIterator[dict]:
+    """Manage server startup and shutdown lifecycle."""
+    try:
+        # Initialize resources on startup
+        await db.connect()
+        yield {"db": db}
+    finally:
+        # Clean up on shutdown
+        await db.disconnect()
+
+# Pass lifespan to server
+server = Server("example-server", lifespan=server_lifespan)
+
+# Access lifespan context in handlers
+@server.call_tool()
+async def query_db(name: str, arguments: dict) -> list:
+    ctx = server.request_context
+    db = ctx.lifespan_context["db"]
+    return await db.query(arguments["query"])
+```
+
+The lifespan API provides:
+- A way to initialize resources when the server starts and clean them up when it stops
+- Access to initialized resources through the request context in handlers
+- Type-safe context passing between lifespan and request handlers
 
 ```python
 from mcp.server.lowlevel import Server, NotificationOptions

src/mcp/server/fastmcp/server.py

@@ -3,8 +3,13 @@
 import inspect
 import json
 import re
+from collections.abc import AsyncIterator
+from contextlib import (
+    AbstractAsyncContextManager,
+    asynccontextmanager,
+)
 from itertools import chain
-from typing import Any, Callable, Literal, Sequence
+from typing import Any, Callable, Generic, Literal, Sequence
 
 import anyio
 import pydantic_core
@@ -19,8 +24,16 @@
 from mcp.server.fastmcp.tools import ToolManager
 from mcp.server.fastmcp.utilities.logging import configure_logging, get_logger
 from mcp.server.fastmcp.utilities.types import Image
-from mcp.server.lowlevel import Server as MCPServer
 from mcp.server.lowlevel.helper_types import ReadResourceContents
+from mcp.server.lowlevel.server import (
+    LifespanResultT,
+)
+from mcp.server.lowlevel.server import (
+    Server as MCPServer,
+)
+from mcp.server.lowlevel.server import (
+    lifespan as default_lifespan,
+)
 from mcp.server.sse import SseServerTransport
 from mcp.server.stdio import stdio_server
 from mcp.shared.context import RequestContext
@@ -50,7 +63,7 @@
 logger = get_logger(__name__)
 
 
-class Settings(BaseSettings):
+class Settings(BaseSettings, Generic[LifespanResultT]):
     """FastMCP server settings.
 
     All settings can be configured via environment variables with the prefix FASTMCP_.
@@ -85,13 +98,36 @@ class Settings(BaseSettings):
         description="List of dependencies to install in the server environment",
     )
 
+    lifespan: (
+        Callable[["FastMCP"], AbstractAsyncContextManager[LifespanResultT]] | None
+    ) = Field(None, description="Lifespan context manager")
+
+
+def lifespan_wrapper(
+    app: "FastMCP",
+    lifespan: Callable[["FastMCP"], AbstractAsyncContextManager[LifespanResultT]],
+) -> Callable[[MCPServer], AbstractAsyncContextManager[object]]:
+    @asynccontextmanager
+    async def wrap(s: MCPServer) -> AsyncIterator[object]:
+        async with lifespan(app) as context:
+            yield context
+
+    return wrap
+
 
 class FastMCP:
     def __init__(
         self, name: str | None = None, instructions: str | None = None, **settings: Any
     ):
         self.settings = Settings(**settings)
-        self._mcp_server = MCPServer(name=name or "FastMCP", instructions=instructions)
+
+        self._mcp_server = MCPServer(
+            name=name or "FastMCP",
+            instructions=instructions,
+            lifespan=lifespan_wrapper(self, self.settings.lifespan)
+            if self.settings.lifespan
+            else default_lifespan,
+        )
         self._tool_manager = ToolManager(
             warn_on_duplicate_tools=self.settings.warn_on_duplicate_tools
         )

src/mcp/server/lowlevel/server.py

@@ -68,7 +68,8 @@ async def main():
 import logging
 import warnings
 from collections.abc import Awaitable, Callable
-from typing import Any, Sequence
+from contextlib import AbstractAsyncContextManager, asynccontextmanager
+from typing import Any, AsyncIterator, Generic, Sequence, TypeVar
 
 from anyio.streams.memory import MemoryObjectReceiveStream, MemoryObjectSendStream
 from pydantic import AnyUrl
@@ -84,7 +85,10 @@ async def main():
 
 logger = logging.getLogger(__name__)
 
-request_ctx: contextvars.ContextVar[RequestContext[ServerSession]] = (
+LifespanResultT = TypeVar("LifespanResultT")
+
+# This will be properly typed in each Server instance's context
+request_ctx: contextvars.ContextVar[RequestContext[ServerSession, Any]] = (
     contextvars.ContextVar("request_ctx")
 )
 
@@ -101,13 +105,33 @@ def __init__(
         self.tools_changed = tools_changed
 
 
-class Server:
+@asynccontextmanager
+async def lifespan(server: "Server") -> AsyncIterator[object]:
+    """Default lifespan context manager that does nothing.
+
+    Args:
+        server: The server instance this lifespan is managing
+
+    Returns:
+        An empty context object
+    """
+    yield {}
+
+
+class Server(Generic[LifespanResultT]):
     def __init__(
-        self, name: str, version: str | None = None, instructions: str | None = None
+        self,
+        name: str,
+        version: str | None = None,
+        instructions: str | None = None,
+        lifespan: Callable[
+            ["Server"], AbstractAsyncContextManager[LifespanResultT]
+        ] = lifespan,
     ):
         self.name = name
         self.version = version
         self.instructions = instructions
+        self.lifespan = lifespan
         self.request_handlers: dict[
             type, Callable[..., Awaitable[types.ServerResult]]
         ] = {
@@ -188,7 +212,7 @@ def get_capabilities(
         )
 
     @property
-    def request_context(self) -> RequestContext[ServerSession]:
+    def request_context(self) -> RequestContext[ServerSession, LifespanResultT]:
         """If called outside of a request context, this will raise a LookupError."""
         return request_ctx.get()
 
@@ -446,9 +470,14 @@ async def run(
         raise_exceptions: bool = False,
     ):
         with warnings.catch_warnings(record=True) as w:
-            async with ServerSession(
-                read_stream, write_stream, initialization_options
-            ) as session:
+            from contextlib import AsyncExitStack
+
+            async with AsyncExitStack() as stack:
+                lifespan_context = await stack.enter_async_context(self.lifespan(self))
+                session = await stack.enter_async_context(
+                    ServerSession(read_stream, write_stream, initialization_options)
+                )
+
                 async for message in session.incoming_messages:
                     logger.debug(f"Received message: {message}")
 
@@ -460,21 +489,28 @@ async def run(
                         ):
                             with responder:
                                 await self._handle_request(
-                                    message, req, session, raise_exceptions
+                                    message,
+                                    req,
+                                    session,
+                                    lifespan_context,
+                                    raise_exceptions,
                                 )
                         case types.ClientNotification(root=notify):
                             await self._handle_notification(notify)
 
                     for warning in w:
                         logger.info(
-                            f"Warning: {warning.category.__name__}: {warning.message}"
+                            "Warning: %s: %s",
+                            warning.category.__name__,
+                            warning.message,
                         )
 
     async def _handle_request(
         self,
         message: RequestResponder,
         req: Any,
         session: ServerSession,
+        lifespan_context: LifespanResultT,
         raise_exceptions: bool,
     ):
         logger.info(f"Processing request of type {type(req).__name__}")
@@ -491,6 +527,7 @@ async def _handle_request(
                         message.request_id,
                         message.request_meta,
                         session,
+                        lifespan_context,
                     )
                 )
                 response = await handler(req)

src/mcp/shared/context.py

@@ -5,10 +5,12 @@
 from mcp.types import RequestId, RequestParams
 
 SessionT = TypeVar("SessionT", bound=BaseSession)
+LifespanContextT = TypeVar("LifespanContextT")
 
 
 @dataclass
-class RequestContext(Generic[SessionT]):
+class RequestContext(Generic[SessionT, LifespanContextT]):
     request_id: RequestId
     meta: RequestParams.Meta | None
     session: SessionT
+    lifespan_context: LifespanContextT

tests/issues/test_176_progress_token.py

@@ -20,7 +20,10 @@ async def test_progress_token_zero_first_call():
     mock_meta.progressToken = 0  # This is the key test case - token is 0
 
     request_context = RequestContext(
-        request_id="test-request", session=mock_session, meta=mock_meta
+        request_id="test-request",
+        session=mock_session,
+        meta=mock_meta,
+        lifespan_context=None,
     )
 
     # Create context with our mocks

tests/server/fastmcp/test_func_metadata.py

@@ -236,7 +236,7 @@ async def check_call(args):
 
 def test_complex_function_json_schema():
     """Test JSON schema generation for complex function arguments.
-    
+
     Note: Different versions of pydantic output slightly different
     JSON Schema formats for model fields with defaults. The format changed in 2.9.0:
 
@@ -245,34 +245,34 @@ def test_complex_function_json_schema():
          "allOf": [{"$ref": "#/$defs/Model"}],
          "default": {}
        }
-       
+
     2. Since 2.9.0:
        {
          "$ref": "#/$defs/Model",
          "default": {}
        }
-       
+
     Both formats are valid and functionally equivalent. This test accepts either format
     to ensure compatibility across our supported pydantic versions.
-    
+
     This change in format does not affect runtime behavior since:
     1. Both schemas validate the same way
     2. The actual model classes and validation logic are unchanged
     3. func_metadata uses model_validate/model_dump, not the schema directly
     """
     meta = func_metadata(complex_arguments_fn)
     actual_schema = meta.arg_model.model_json_schema()
-    
+
     # Create a copy of the actual schema to normalize
     normalized_schema = actual_schema.copy()
-    
+
     # Normalize the my_model_a_with_default field to handle both pydantic formats
-    if 'allOf' in actual_schema['properties']['my_model_a_with_default']:
-        normalized_schema['properties']['my_model_a_with_default'] = {
-            '$ref': '#/$defs/SomeInputModelA',
-            'default': {}
+    if "allOf" in actual_schema["properties"]["my_model_a_with_default"]:
+        normalized_schema["properties"]["my_model_a_with_default"] = {
+            "$ref": "#/$defs/SomeInputModelA",
+            "default": {},
         }
-    
+
     assert normalized_schema == {
         "$defs": {
             "InnerModel": {