Retire wording tied to pre-2.0 milestones (#3211)

maxisbey · web-flow · commit b31ddf37ed77 · 2026-07-29T14:54:23.000+01:00
diff --git a/docs/advanced/middleware.md b/docs/advanced/middleware.md
@@ -5,8 +5,8 @@ A **middleware** is one async function that wraps every message your server rece
 You write it as `async (ctx, call_next)` and append it to `server.middleware`. That is the whole API.
 
 !!! warning
-    The middleware list is marked **provisional** in the source. The signature and semantics are
-    expected to change before v2 is final. Use it to *observe* (timing, logging, tracing) and to
+    The middleware list is marked **provisional** in the source: its signature and semantics may
+    change in a 2.x minor release. Use it to *observe* (timing, logging, tracing) and to
     *refuse* messages; do not make it the foundation your server stands on.
 
 `MCPServer` takes the list at construction (`MCPServer(name, middleware=[...])`) and exposes it as
diff --git a/examples/stories/legacy_routing/README.md b/examples/stories/legacy_routing/README.md
@@ -94,8 +94,7 @@ eras need different auth, rate limits, or scaling.
 - DNS-rebinding protection is on by default; the harness disables it
   (`NO_DNS_REBIND`) because the in-process httpx2 client sends no `Origin`.
   Drop the kwarg for a real deployment.
-- `mcp.shared.inbound` is a deep import path — a shorter re-export is planned
-  before beta.
+- `mcp.shared.inbound` is a deep import path; there is no shorter re-export.
 
 ## Spec
 
diff --git a/examples/stories/middleware/README.md b/examples/stories/middleware/README.md
@@ -31,12 +31,13 @@ uv run python -m stories.middleware.client --http
 
 ## Caveats
 
-- **Lowlevel-only.** `Server.middleware` on `mcp.server.lowlevel.Server` is the
-  one public hook; `MCPServer` has no public accessor for it yet (a
-  `MCPServer.middleware` accessor is planned before beta).
+- **One list, two accessors.** `Server.middleware` on
+  `mcp.server.lowlevel.Server` is the hook this story uses; `MCPServer`
+  exposes the same list as `MCPServer.middleware` (or takes it at
+  construction as `MCPServer(name, middleware=[...])`).
 - The middleware signature is **provisional** (see the TODO in
-  `src/mcp/server/lowlevel/server.py`): it tightens to a covariant `Context[L]`
-  and gains an outbound seam before v2 final.
+  `src/mcp/server/lowlevel/server.py`): it may change in a 2.x minor release,
+  tightening to a covariant `Context[L]` and gaining an outbound seam.
 - `ServerMiddleware` / `CallNext` / `HandlerResult` are imported from
   `mcp.server.context` (helper tier); not re-exported at `mcp.server.lowlevel`.
 - Do **not** `await ctx.session.send_request(...)` while wrapping `initialize`
diff --git a/examples/stories/middleware/server.py b/examples/stories/middleware/server.py
@@ -1,7 +1,7 @@
 """Dispatch-layer middleware: `Server.middleware` is the public hook.
 
-A lowlevel-only story: `MCPServer` has no public middleware accessor yet, so the
-one supported registration point is the `middleware` list on `lowlevel.Server`.
+This story registers on the lowlevel `Server`; `MCPServer` exposes the same
+list as `MCPServer.middleware`, so the recipe carries over unchanged.
 """
 
 import json
diff --git a/examples/stories/roots/README.md b/examples/stories/roots/README.md
@@ -3,7 +3,6 @@
 > **Deprecated** in the 2026-07-28 protocol (SEP-2577); functional through the
 > deprecation window. Migration: accept directory paths as ordinary tool
 > parameters or resource URIs instead of relying on `roots/list`.
-> TODO(maxisbey): revisit before beta.
 
 The client passes a `list_roots_callback` returning the filesystem locations it
 is willing to expose; a server tool calls `ctx.session.list_roots()` mid-request
diff --git a/examples/stories/sampling/README.md b/examples/stories/sampling/README.md
@@ -3,7 +3,6 @@
 > **Deprecated** in the 2026-07-28 protocol (SEP-2577); functional through the
 > deprecation window. Migration: call your LLM provider directly from the
 > server instead of requesting completions through the client.
-> TODO(maxisbey): revisit before beta.
 
 A tool that asks the **client's** LLM for a completion mid-call — the inverted
 MCP direction. The server holds no model API key; it awaits
diff --git a/examples/stories/serve_one/README.md b/examples/stories/serve_one/README.md
@@ -34,15 +34,15 @@ uv run python -m stories.serve_one.client
 ## Caveats
 
 - **Deep imports** — `serve_one`, `serve_connection`, and `Connection` are only
-  reachable at `mcp.server.runner` / `mcp.server.connection` today; a shorter
-  `mcp.server.*` re-export is tracked for beta.
+  reachable at `mcp.server.runner` / `mcp.server.connection`; there is no
+  shorter `mcp.server.*` re-export.
 - **Lowlevel-only.** The drivers take a `lowlevel.Server` and `MCPServer` has
   no public accessor for its underlying one (`_lowlevel_server` is private), so
   there is no `MCPServer`-tier variant of this story. Build the lowlevel
   `Server` directly until that accessor lands.
 - **No public `DispatchContext`** — `SingleExchangeContext` is hand-rolled
-  boilerplate; a public helper (or a `serve_one` overload that builds one) is
-  tracked for beta.
+  boilerplate; there is no public helper (or `serve_one` overload) that builds
+  one.
 - **Lifespan** — the transport entry enters `server.lifespan(server)` **once**
   and threads `lifespan_state` to every `handle_one()` call; never enter it
   per-request.
diff --git a/examples/stories/sse_polling/README.md b/examples/stories/sse_polling/README.md
@@ -4,7 +4,7 @@
 > the sessionful transport are removed in the 2026-07-28 protocol (SEP-2575)
 > with no modern-era equivalent; the closest 2026-era pattern is client-side
 > reconnection over a persisted `DiscoverResult` —
-> [`reconnect/`](../reconnect/). TODO(maxisbey): revisit before beta.
+> [`reconnect/`](../reconnect/).
 
 SEP-1699 server-initiated SSE disconnection with `Last-Event-ID` replay. The
 server's `EventStore` stamps every SSE event with an ID and opens each response
diff --git a/examples/stories/starlette_mount/README.md b/examples/stories/starlette_mount/README.md
@@ -45,8 +45,7 @@ kill "$SERVER_PID"
   no `Origin` header. Remove it (or configure allowed hosts) for a real
   deployment.
 - The parent-lifespan dance is a known SDK ergonomics gap (other SDKs mount
-  with no extra ceremony); tracked for the beta reshape. The recipe shown here
-  is what works today.
+  with no extra ceremony). The recipe shown here is what works today.
 
 ## Spec
 
diff --git a/examples/stories/streaming/README.md b/examples/stories/streaming/README.md
@@ -55,7 +55,7 @@ uv run python -m stories.streaming.client --http --server server_lowlevel
   through the deprecation window. Migration: write to stderr or emit
   OpenTelemetry instead of `notifications/message`. It is shown here because
   servers still need to support 2025-era clients during that window. Progress
-  and cancellation are **not** deprecated. TODO(maxisbey): revisit before beta.
+  and cancellation are **not** deprecated.
 - A cancelled request is not answered: no response follows
   `notifications/cancelled`. (The 2025-era streamable HTTP transport is the one
   exception - its wire ends a request only with a response, so it terminates
diff --git a/src/mcp/server/lowlevel/server.py b/src/mcp/server/lowlevel/server.py
@@ -433,9 +433,9 @@ def __init__(
         # `OpenTelemetryMiddleware` ships on by default so every server emits a
         # SERVER span per message; it is a no-op until an OTel exporter is
         # installed. Drop it from this list to opt out.
-        # TODO(L54): provisional - signature and semantics change with the
-        # Context/middleware rework (covariant `Context[L]`, outbound seam) before
-        # v2 final.
+        # TODO(L54): provisional - signature and semantics may change in a 2.x
+        # minor release with the Context/middleware rework (covariant
+        # `Context[L]`, outbound seam).
         self.middleware: list[ServerMiddleware[LifespanResultT]] = [OpenTelemetryMiddleware()]
         # SEP-2133 extension settings advertised under `ServerCapabilities.extensions`
         # (identifier -> settings). Higher layers (e.g. `MCPServer(extensions=...)`)
diff --git a/src/mcp/server/mcpserver/server.py b/src/mcp/server/mcpserver/server.py
@@ -267,8 +267,8 @@ def middleware(self) -> list[ServerMiddleware[Any]]:
 
         The same list as the low-level `Server.middleware`: append an
         `async (ctx, call_next)` callable to observe, refuse, or rewrite
-        messages before they reach a handler. Provisional - the signature is
-        expected to change before v2 is final; see the middleware guide.
+        messages before they reach a handler. Provisional - the signature may
+        change in a 2.x minor release; see the middleware guide.
         """
         return self._lowlevel_server.middleware
 
diff --git a/src/mcp/shared/dispatcher.py b/src/mcp/shared/dispatcher.py
@@ -250,7 +250,8 @@ class Dispatcher(Outbound, Protocol[TransportT_co]):
     Implementations own correlation of outbound requests to inbound results, the
     receive loop, per-request concurrency, and cancellation/progress wiring.
 
-    The lifecycle surface is provisional; `run()` may change before v2 stable.
+    The lifecycle surface is provisional; `run()` may change in a 2.x minor
+    release.
     """
 
     async def run(
diff --git a/tests/interaction/_requirements.py b/tests/interaction/_requirements.py
@@ -660,7 +660,7 @@ def __post_init__(self) -> None:
                 "The dispatcher drops null-id error responses with a debug log; in v1, JSONRPCError.id was "
                 "non-nullable, so a null-id error response failed transport validation and the resulting "
                 "ValidationError was surfaced to message_handler as an exception. A typed fault channel "
-                "restoring visibility is planned before v2 stable."
+                "restoring visibility is planned."
             ),
         ),
         deferred=(
