Slim ServerMiddleware to (ctx, call_next) and add OpenTelemetryMiddleware

docs/migration.md

@@ -720,7 +720,7 @@ ctx: ClientRequestContext
 server_ctx: ServerRequestContext[LifespanContextT, RequestT]
 ```
 
-`ServerRequestContext` is now a standalone dataclass — it no longer subclasses `RequestContext[ServerSession]`. It carries the same fields (`session`, `request_id`, `meta`, `lifespan_context`, `request`, `close_sse_stream`, `close_standalone_sse_stream`) plus a new `protocol_version: str` field, so handler code is unaffected, but `isinstance(ctx, RequestContext)` checks and `RequestContext[ServerSession]` annotations need updating to `ServerRequestContext`.
+`ServerRequestContext` is now a standalone dataclass — it no longer subclasses `RequestContext[ServerSession]`. It carries the same fields (`session`, `request_id`, `meta`, `lifespan_context`, `request`, `close_sse_stream`, `close_standalone_sse_stream`) plus new `protocol_version: str`, `method: str`, and raw `params: Mapping[str, Any] | None` fields (the last two let middleware read and rewrite the inbound message), so handler code is unaffected, but `isinstance(ctx, RequestContext)` checks and `RequestContext[ServerSession]` annotations need updating to `ServerRequestContext`.
 
 The high-level `Context` class (injected into `@mcp.tool()` etc.) similarly dropped its `ServerSessionT` parameter: `Context[ServerSessionT, LifespanContextT, RequestT]` → `Context[LifespanContextT, RequestT]`. Both remaining parameters have defaults, so bare `Context` is usually sufficient:
 
@@ -935,27 +935,24 @@ server.add_notification_handler("notifications/custom", MyNotifyParams, my_notif
 These were private, but some users subclassed `Server` and overrode them to intercept requests. Use middleware instead:
 
 ```python
-from collections.abc import Mapping
 from typing import Any
 
 from mcp.server import Server, ServerRequestContext
 from mcp.server.context import CallNext, HandlerResult
 
 
-async def logging_middleware(
-    ctx: ServerRequestContext[Any, Any], method: str, params: Mapping[str, Any] | None, call_next: CallNext
-) -> HandlerResult:
-    print(f"handling {method}")
-    result = await call_next()
-    print(f"done {method}")
+async def logging_middleware(ctx: ServerRequestContext[Any, Any], call_next: CallNext) -> HandlerResult:
+    print(f"handling {ctx.method}")
+    result = await call_next(ctx)
+    print(f"done {ctx.method}")
     return result
 
 
 server = Server("my-server", on_call_tool=...)
 server.middleware.append(logging_middleware)
 ```
 
-Middleware runs before params validation, so `params` is the raw inbound mapping (or `None`), and it also wraps unknown methods.
+The method and the raw inbound params are `ctx.method` and `ctx.params` (`params` is `None` when the message carries none). Middleware runs before params validation and also wraps unknown methods. To rewrite the method or params before the handler runs, pass an adjusted context through: `await call_next(replace(ctx, params=...))`.
 
 ### Lowlevel `Server.run(raise_exceptions=True)`: transport errors no longer re-raised
 

src/mcp/server/_otel.py

@@ -0,0 +1,51 @@
+from __future__ import annotations
+
+from typing import Any
+
+from opentelemetry.trace import SpanKind, StatusCode
+from pydantic import ValidationError
+
+from mcp.server.context import CallNext, HandlerResult, ServerMiddleware, ServerRequestContext
+from mcp.shared._otel import extract_trace_context, otel_span
+from mcp.shared.exceptions import MCPError
+
+
+class OpenTelemetryMiddleware(ServerMiddleware[Any]):
+    """Context-tier middleware that wraps each inbound message in an OpenTelemetry span.
+
+    Span name `"MCP handle <method> [<target>]"`, `mcp.method.name` attribute, W3C
+    trace context extracted from `params._meta` (SEP-414), and an ERROR status if
+    the handler raises. Requests and notifications both get a span;
+    `jsonrpc.request.id` is set only when `ctx.request_id` is present (notifications
+    have none).
+    """
+
+    async def __call__(self, ctx: ServerRequestContext[Any, Any], call_next: CallNext) -> HandlerResult:
+        name = ctx.params.get("name") if ctx.params else None
+        target = name if isinstance(name, str) else None
+
+        attributes: dict[str, Any] = {"mcp.method.name": ctx.method}
+        if ctx.request_id is not None:
+            attributes["jsonrpc.request.id"] = str(ctx.request_id)
+
+        with otel_span(
+            name=f"MCP handle {ctx.method}{f' {target}' if target else ''}",
+            kind=SpanKind.SERVER,
+            attributes=attributes,
+            context=extract_trace_context(ctx.meta or {}),
+            record_exception=False,
+            set_status_on_exception=False,
+        ) as span:
+            try:
+                return await call_next(ctx)
+            except MCPError as e:
+                span.set_status(StatusCode.ERROR, e.error.message)
+                raise
+            except ValidationError:
+                # Mirror the sanitized wire response; pydantic messages carry client input.
+                span.set_status(StatusCode.ERROR, "Invalid request parameters")
+                raise
+            except Exception as e:
+                span.record_exception(e)
+                span.set_status(StatusCode.ERROR, str(e))
+                raise

src/mcp/server/context.py

@@ -15,7 +15,7 @@
 from mcp.shared.transport_context import TransportContext
 from mcp.types import LoggingLevel, RequestId, RequestParamsMeta
 
-# Invariant: parameterizes a mutable dataclass field; dict default matches the default lifespan.
+# Invariant: parametrizes a mutable dataclass field; dict default matches the default lifespan.
 LifespanContextT = TypeVar("LifespanContextT", default=dict[str, Any])
 RequestT = TypeVar("RequestT", default=Any)
 
@@ -33,6 +33,8 @@ class ServerRequestContext(Generic[LifespanContextT, RequestT]):
     session: ServerSession
     lifespan_context: LifespanContextT
     protocol_version: str
+    method: str
+    params: Mapping[str, Any] | None = None
     request_id: RequestId | None = None
     meta: RequestParamsMeta | None = None
     request: RequestT | None = None
@@ -113,39 +115,41 @@ async def log(self, level: LoggingLevel, data: Any, logger: str | None = None, *
 """What a request handler (or middleware) may return. `ServerRunner` serializes
 all three to a result dict."""
 
-CallNext = Callable[[], Awaitable[HandlerResult]]
+CallNext = Callable[["ServerRequestContext[Any, Any]"], Awaitable[HandlerResult]]
+"""Invokes the rest of the chain. Pass the `ctx` through; rewrite `method` or
+`params` with `dataclasses.replace(ctx, ...)` to alter what the handler sees."""
 
 _MwLifespanT = TypeVar("_MwLifespanT")
 
 
 class ServerMiddleware(Protocol[_MwLifespanT]):
-    """Context-tier middleware: `(ctx, method, params, call_next) -> result`.
+    """Context-tier middleware: `(ctx, call_next) -> result`.
 
     Runs at the top of `ServerRunner._on_request` / `_on_notify` after `ctx`
     is built but before any validation, lookup, or handshake. Wraps every
     inbound request and notification: `initialize`, the pre-init gate,
     `METHOD_NOT_FOUND`, params validation, the handler call, and
-    `notifications/initialized` all run inside `call_next()`.
+    `notifications/initialized` all run inside `call_next(ctx)`.
     `notifications/cancelled` is observed too; the dispatcher applies the
     cancellation itself, then forwards the notification. A request-side
     failure reaches the middleware as a raised `MCPError` (or
     `ValidationError` for malformed params) so observation/logging middleware
     can record it. Listed outermost-first on `Server.middleware`.
 
+    The method and the raw inbound params are `ctx.method` and `ctx.params` (no
+    model validation has happened yet). To rewrite either before the handler
+    runs, pass an adjusted context: `await call_next(replace(ctx, params=...))`.
     `ctx.request_id is None` distinguishes a notification from a request. For
-    notifications `call_next()` returns `None` (a dropped or unhandled
+    notifications `call_next(ctx)` returns `None` (a dropped or unhandled
     notification also returns `None`) and the middleware's own return value is
     discarded.
 
-    `params` is the raw inbound mapping (no model validation has happened
-    yet). For typed inspection, validate against the model the middleware
-    expects.
-
-    Warning: `initialize` is handled inline - the dispatcher does not read
-    further inbound messages until the middleware chain returns. Awaiting a
-    server-to-client request (`ctx.session.send_request`, `send_ping`, ...)
-    while handling `initialize` therefore deadlocks the connection: the
-    response can never be dequeued. Send-and-forget notifications are safe.
+    !!! warning
+        `initialize` is handled inline - the dispatcher does not read
+        further inbound messages until the middleware chain returns. Awaiting a
+        server-to-client request (`ctx.session.send_request`, `send_ping`, ...)
+        while handling `initialize` therefore deadlocks the connection: the
+        response can never be dequeued. Send-and-forget notifications are safe.
 
     `Server[L].middleware` holds `ServerMiddleware[L]`, so an app-specific
     middleware sees `ctx.lifespan_context: L`. While the context is the
@@ -162,7 +166,5 @@ class ServerMiddleware(Protocol[_MwLifespanT]):
     async def __call__(
         self,
         ctx: ServerRequestContext[_MwLifespanT, Any],
-        method: str,
-        params: Mapping[str, Any] | None,
         call_next: CallNext,
     ) -> HandlerResult: ...

src/mcp/server/lowlevel/server.py

@@ -217,7 +217,7 @@ def __init__(
         self._session_manager: StreamableHTTPSessionManager | None = None
         # Context-tier middleware: wraps every inbound request (including
         # `initialize`, lookup, validation, handler) with
-        # `(ctx, method, params, call_next)`. Applied in `ServerRunner._on_request`.
+        # `(ctx, call_next)`. Applied in `ServerRunner._on_request`.
         # TODO(maxisbey): provisional - signature and semantics change with the
         # Context/middleware rework (covariant `Context[L]`, outbound seam) before
         # v2 final.

src/mcp/server/runner.py

@@ -16,7 +16,7 @@
 from __future__ import annotations
 
 import logging
-from collections.abc import Mapping
+from collections.abc import Awaitable, Mapping
 from dataclasses import dataclass, field
 from functools import partial, reduce
 from typing import TYPE_CHECKING, Any, Generic, cast
@@ -103,7 +103,7 @@
     return "2025-11-25"
 
 
-def otel_middleware(next_on_request: OnRequest) -> OnRequest:
+def otel_middleware(call_next: OnRequest) -> OnRequest:
     """Dispatch-tier middleware that wraps each request in an OpenTelemetry span.
 
     Mirrors the span shape of the existing `Server._handle_request`: span name
@@ -139,7 +139,7 @@
             set_status_on_exception=False,
         ) as span:
             try:
-                return await next_on_request(dctx, method, params)
+                return await call_next(dctx, method, params)
             except MCPError as e:
                 span.set_status(StatusCode.ERROR, e.error.message)
                 raise
@@ -169,6 +169,14 @@
     raise TypeError(f"handler returned {type(result).__name__}; expected BaseModel, dict, or None")
 
 
+def _apply_middleware(
+    mw: ServerMiddleware[Any], call_next: CallNext, ctx: ServerRequestContext[Any, Any]
+) -> Awaitable[HandlerResult]:
+    """Adapt one middleware to the `CallNext` shape: bind `call_next`, take
+    `ctx` at call time so a rewritten context flows down the chain."""
+    return mw(ctx, call_next)
+
+
 @dataclass
 class ServerRunner(Generic[LifespanT]):
     """Per-connection orchestrator. One instance per client connection."""
@@ -244,15 +252,18 @@
     ) -> dict[str, Any]:
         meta = _extract_meta(params)
         version = _resolve_protocol_version(self.connection.protocol_version, meta, dctx.message_metadata)
-        ctx = self._make_context(dctx, meta, version)
+        ctx = self._make_context(dctx, method, params, meta, version)
         is_spec_method = method in _methods.SPEC_CLIENT_METHODS
 
-        async def _inner() -> HandlerResult:
+        async def _inner(ctx: ServerRequestContext[LifespanT, Any]) -> HandlerResult:
+            # Read method/params off `ctx` so a middleware that rewrote them via
+            # `call_next(replace(ctx, ...))` reaches lookup and the handler.
+            method, params = ctx.method, ctx.params
             # Pinned compat: spec methods are surface-validated before lookup,
             # so malformed params are INVALID_PARAMS even with no handler
             # registered. Custom methods miss the monolith map and fall through
             # to `entry.params_type` exactly as before.
-            if is_spec_method:
+            if method in _methods.SPEC_CLIENT_METHODS:
                 try:
                     _methods.validate_client_request(method, version, params)
                 except KeyError:
@@ -279,14 +290,14 @@
             result = await entry.handler(ctx, typed_params)
             if isinstance(result, ErrorData):
                 # Raise inside the chain so middleware observes the failure.
                 raise MCPError.from_error_data(result)
             return result
 
-        call = self._compose_server_middleware(ctx, method, params, _inner)
-        result = _dump_result(await call())
+        call = self._compose_server_middleware(_inner)
+        result = _dump_result(await call(ctx))
         # TODO: reject resultType values outside {"complete", "input_required"} unless the
         # corresponding extension is in this request's _meta clientCapabilities.extensions; the
         # explicit MUST-reject is client-side (basic/index.mdx ResultType), this enforces it proactively.
         if is_spec_method:
             try:
                 result = _methods.serialize_server_result(method, version, result)
@@ -313,9 +324,10 @@
     ) -> None:
         meta = _extract_meta(params)
         version = _resolve_protocol_version(self.connection.protocol_version, meta, dctx.message_metadata)
-        ctx = self._make_context(dctx, meta, version)
+        ctx = self._make_context(dctx, method, params, meta, version)
 
-        async def _inner() -> None:
+        async def _inner(ctx: ServerRequestContext[LifespanT, Any]) -> None:
+            method, params = ctx.method, ctx.params
             if method in _methods.SPEC_CLIENT_NOTIFICATION_METHODS:
                 try:
                     _methods.validate_client_notification(method, version, params)
@@ -345,33 +357,33 @@
                 return
             await entry.handler(ctx, typed_params)
 
-        call = self._compose_server_middleware(ctx, method, params, _inner)
+        call = self._compose_server_middleware(_inner)
         try:
-            await call()
+            await call(ctx)
         except Exception:
             # A crashing handler must not cancel the dispatcher's task group;
             # middleware saw the raise out of call_next() first.
             logger.exception("notification handler for %r raised", method)
 
-    def _compose_server_middleware(
-        self,
-        ctx: ServerRequestContext[LifespanT, Any],
-        method: str,
-        params: Mapping[str, Any] | None,
-        inner: CallNext,
-    ) -> CallNext:
+    def _compose_server_middleware(self, inner: CallNext) -> CallNext:
         """Wrap `inner` in `Server.middleware`, outermost-first.
 
         Shared by `_on_request` and `_on_notify` so the same middleware chain
-        observes every inbound message.
+        observes every inbound message. The composed callable takes the `ctx`
+        at call time, so a middleware can rewrite it for the rest of the chain.
         """
-        call = inner
+        call: CallNext = inner
         for mw in reversed(self.server.middleware):
-            call = partial(mw, ctx, method, params, call)
+            call = partial(_apply_middleware, mw, call)
         return call
 
     def _make_context(
-        self, dctx: DispatchContext[TransportContext], meta: RequestParamsMeta | None, protocol_version: str
+        self,
+        dctx: DispatchContext[TransportContext],
+        method: str,
+        params: Mapping[str, Any] | None,
+        meta: RequestParamsMeta | None,
+        protocol_version: str,
     ) -> ServerRequestContext[LifespanT, Any]:
         # TODO(maxisbey): remove for Context rework. Reads the SHTTP per-request
         # data off the raw `dctx.message_metadata` carrier; replace with the
@@ -386,6 +398,8 @@
         return ServerRequestContext(
             session=self.session,
             lifespan_context=self.lifespan_state,
+            method=method,
+            params=params,
             request_id=dctx.request_id,
             meta=meta,
             protocol_version=protocol_version,

src/mcp/shared/_otel.py

@@ -2,13 +2,14 @@
 
 from __future__ import annotations
 
-from collections.abc import Iterator
+from collections.abc import Generator, Mapping
 from contextlib import contextmanager
 from typing import Any
 
 from opentelemetry.context import Context
 from opentelemetry.propagate import extract, inject
 from opentelemetry.trace import SpanKind, get_tracer
+from opentelemetry.trace.span import Span
 
 _tracer = get_tracer("mcp-python-sdk")
 
@@ -22,7 +23,7 @@ def otel_span(
     context: Context | None = None,
     record_exception: bool = True,
     set_status_on_exception: bool = True,
-) -> Iterator[Any]:
+) -> Generator[Span]:
     """Create an OTel span."""
     with _tracer.start_as_current_span(
         name,
@@ -40,13 +41,10 @@ def inject_trace_context(meta: dict[str, Any]) -> None:
     inject(meta)
 
 
-def extract_trace_context(meta: dict[str, Any]) -> Context | None:
-    """Extract W3C trace context from a `_meta` dict.
-
-    Returns `None` when the carrier is malformed; telemetry parsing must
-    never fail the request it annotates.
-    """
+def extract_trace_context(meta: Mapping[str, Any]) -> Context:
+    """Extract W3C trace context from a `_meta` dict."""
     try:
         return extract(meta)
-    except (TypeError, ValueError):
-        return None
+    except (ValueError, TypeError):
+        # If the traceparent is malformed, degrade to no parent rather than failing the request.
+        return Context()

tests/issues/test_176_progress_token.py

@@ -19,6 +19,7 @@ async def test_progress_token_zero_first_call():
     request_context = ServerRequestContext(
         request_id="test-request",
         session=mock_session,
+        method="tools/call",
         meta={"progress_token": 0},
         lifespan_context=None,
         protocol_version="2025-11-25",

tests/server/mcpserver/test_server.py

@@ -1528,6 +1528,7 @@ async def test_report_progress_passes_related_request_id():
     request_context = ServerRequestContext(
         request_id="req-abc-123",
         session=mock_session,
+        method="tools/call",
         meta={"progress_token": "tok-1"},
         lifespan_context=None,
         protocol_version="2025-11-25",