fix: address Copilot review round 3 on #2318

- fetchTracking: the close report is guarded so a throwing listener
  cannot become an unhandled rejection off the discarded watcher; a
  bare CR that is the last byte of the stream terminates its event; a
  moving event count is coalesced to one report per interval (default
  250ms, 0 in tests), with the open and the close still immediate.
- MessageTrackingTransport rolls back a request whose signal was already
  aborted before the send began, and suppresses the rollback only for an
  abort that lands during the send.
- useTickingClock takes an `enabled` flag; Connection Info installs no
  interval when there is no activity section.
- The diagnostics-callbacks test justifies its private-seam cast.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01APKSYHGxXBUpnR7ErFiLJn
Signed-off-by: cliffhall <cliff@futurescale.com>

clients/web/src/components/groups/ConnectionInfoContent/ConnectionInfoContent.tsx

@@ -337,8 +337,13 @@ export function ConnectionInfoContent({
     : SERVER_INFO_NOT_REPORTED_LABEL;
 
   // The activity rows' clock: the snapshot's own on first paint (pure), then
-  // the wall clock once a second so "sent 5s ago" keeps counting.
-  const now = useTickingClock(diagnostics?.capturedAt ?? 0);
+  // the wall clock once a second so "sent 5s ago" keeps counting. No timer
+  // at all when there is no activity section to tick.
+  const now = useTickingClock(
+    diagnostics?.capturedAt ?? 0,
+    1000,
+    diagnostics !== undefined,
+  );
 
   const serverCaps = getServerCapabilityEntries(capabilities, protocolEra);
   const clientCaps = getCapabilityEntries(

clients/web/src/hooks/useTickingClock.test.ts

@@ -47,6 +47,28 @@ describe("useTickingClock", () => {
     expect(result.current).toBe(5_002_500);
   });
 
+  it("installs no timer and returns the seed when disabled", () => {
+    vi.useFakeTimers();
+    vi.setSystemTime(5_000_000);
+    const setSpy = vi.spyOn(globalThis, "setInterval");
+    const { result, rerender } = renderHook(
+      ({ enabled }) => useTickingClock(1_000_000, 1000, enabled),
+      { initialProps: { enabled: false } },
+    );
+    expect(setSpy).not.toHaveBeenCalled();
+    act(() => {
+      vi.advanceTimersByTime(3000);
+    });
+    expect(result.current).toBe(1_000_000);
+    // Enabling later starts the clock.
+    rerender({ enabled: true });
+    act(() => {
+      vi.advanceTimersByTime(1000);
+    });
+    expect(result.current).toBe(5_004_000);
+    setSpy.mockRestore();
+  });
+
   it("stops ticking on unmount", () => {
     vi.useFakeTimers();
     const clearSpy = vi.spyOn(globalThis, "clearInterval");

clients/web/src/hooks/useTickingClock.ts

@@ -16,12 +16,20 @@ import { useEffect, useState } from "react";
  * A timer is exactly the "synchronize with an external system" case an effect
  * exists for; the state is set only from the interval callback, never in the
  * effect body, so the first paint is not followed by a corrective re-render.
+ * `enabled: false` installs no timer at all and returns the seed.
  */
-export function useTickingClock(seed: number, intervalMs = 1000): number {
+export function useTickingClock(
+  seed: number,
+  intervalMs = 1000,
+  enabled = true,
+): number {
   const [now, setNow] = useState(seed);
   useEffect(() => {
+    // No timer for a consumer with nothing to tick — a component that
+    // renders no durations must not pay for a repeating interval.
+    if (!enabled) return;
     const id = setInterval(() => setNow(Date.now()), intervalMs);
     return () => clearInterval(id);
-  }, [intervalMs]);
+  }, [intervalMs, enabled]);
   return Math.max(now, seed);
 }

clients/web/src/test/core/mcp/fetchTracking.test.ts

@@ -748,6 +748,7 @@ describe("createFetchTracker long-lived stream watching (#2318)", () => {
       {
         trackRequest: (entry) => tracked.push(entry),
         updateStream: (id, state) => updates.push({ id, stream: state }),
+        streamUpdateIntervalMs: 0,
       },
     );
     const response = await fetcher("https://example.com/mcp", {
@@ -800,7 +801,10 @@ describe("createFetchTracker long-lived stream watching (#2318)", () => {
         new Response(server.stream, {
           headers: { "content-type": "application/x-ndjson" },
         })) as typeof fetch,
-      { updateStream: (id, state) => updates.push({ id, stream: state }) },
+      {
+        updateStream: (id, state) => updates.push({ id, stream: state }),
+        streamUpdateIntervalMs: 0,
+      },
     );
     await fetcher("https://example.com/mcp", { method: "GET" });
     server.push('{"jsonrpc":"2.0","method":"a"}\n\n');
@@ -846,7 +850,10 @@ describe("createFetchTracker long-lived stream watching (#2318)", () => {
         new Response(server.stream, {
           headers: { "content-type": "Text/Event-Stream" },
         })) as typeof fetch,
-      { updateStream: (id, state) => updates.push({ id, state }) },
+      {
+        updateStream: (id, state) => updates.push({ id, state }),
+        streamUpdateIntervalMs: 0,
+      },
     );
     await fetcher("https://example.com/mcp", { method: "get" });
     expect(updates).toHaveLength(1);
@@ -867,6 +874,7 @@ describe("createFetchTracker long-lived stream watching (#2318)", () => {
         trackRequest: (entry) => tracked.push(entry),
         updateResponseBody: (_id, body) => bodies.push(body),
         updateStream: (id, state) => updates.push({ id, state }),
+        streamUpdateIntervalMs: 0,
       },
     );
     await fetcher(new Request("https://example.com/mcp", { method: "POST" }));
@@ -876,6 +884,80 @@ describe("createFetchTracker long-lived stream watching (#2318)", () => {
     expect(updates).toEqual([]);
   });
 
+  it("coalesces a moving count to one report per interval, and reports the close at once", async () => {
+    vi.useFakeTimers();
+    try {
+      const server = controlledStream();
+      const updates: Array<{ id: string; stream: FetchStreamState }> = [];
+      const fetcher = createFetchTracker(
+        (async () =>
+          new Response(server.stream, {
+            headers: { "content-type": "text/event-stream" },
+          })) as typeof fetch,
+        {
+          updateStream: (id, state) => updates.push({ id, stream: state }),
+          streamUpdateIntervalMs: 100,
+        },
+      );
+      await fetcher("https://example.com/mcp", { method: "GET" });
+      // The open is immediate.
+      expect(updates.map((u) => u.stream.eventCount)).toEqual([0]);
+      server.push("data: a\n\ndata: b\n\ndata: c\n\n");
+      await vi.advanceTimersByTimeAsync(0);
+      // Three events, no report yet — they are inside one window.
+      expect(updates.map((u) => u.stream.eventCount)).toEqual([0]);
+      await vi.advanceTimersByTimeAsync(100);
+      expect(updates.map((u) => u.stream.eventCount)).toEqual([0, 3]);
+      // A close cancels a pending window and reports immediately.
+      server.push("data: d\n\n");
+      server.end();
+      await vi.advanceTimersByTimeAsync(0);
+      expect(updates.map((u) => u.stream.eventCount)).toEqual([0, 3, 4]);
+      expect(updates.at(-1)!.stream.closedAt).toBeInstanceOf(Date);
+      await vi.advanceTimersByTimeAsync(200);
+      expect(updates).toHaveLength(3);
+    } finally {
+      vi.useRealTimers();
+    }
+  });
+
+  it("counts an event whose bare-CR terminator is the last byte of the stream", async () => {
+    const server = controlledStream();
+    const { updates } = await trackStream(server.stream);
+    server.push("data: x\r\r");
+    await flush();
+    // Held while the stream is open: the CR could be half of a CRLF.
+    expect(updates.map((u) => u.stream.eventCount)).toEqual([0]);
+    server.end();
+    await flush();
+    expect(updates.at(-1)!.stream).toMatchObject({ eventCount: 1 });
+    expect(updates.at(-1)!.stream.closedAt).toBeInstanceOf(Date);
+  });
+
+  it("survives a stream-update listener that throws on the close", async () => {
+    // The watcher's promise is discarded, so a throw from the final report
+    // would otherwise be an unhandled rejection — which fails the whole run.
+    const server = controlledStream();
+    let reports = 0;
+    const fetcher = createFetchTracker(
+      (async () =>
+        new Response(server.stream, {
+          headers: { "content-type": "text/event-stream" },
+        })) as typeof fetch,
+      {
+        updateStream: (_id, state) => {
+          reports += 1;
+          if (state.closedAt) throw new Error("listener exploded");
+        },
+        streamUpdateIntervalMs: 0,
+      },
+    );
+    await fetcher("https://example.com/mcp", { method: "GET" });
+    server.end();
+    await flush();
+    expect(reports).toBe(2);
+  });
+
   it("reports the close when the server ends the stream", async () => {
     const server = controlledStream();
     const { updates, id } = await trackStream(server.stream);
@@ -932,7 +1014,10 @@ describe("createFetchTracker long-lived stream watching (#2318)", () => {
         new Response("data: x\n\n", {
           headers: { "content-type": "text/event-stream" },
         })) as typeof fetch,
-      { updateStream: (id, state) => updates.push({ id, state }) },
+      {
+        updateStream: (id, state) => updates.push({ id, state }),
+        streamUpdateIntervalMs: 0,
+      },
     );
     // A POST SSE reply is bounded and goes down the body-capture path.
     await fetcher("https://example.com/mcp", { method: "POST" });
@@ -950,6 +1035,7 @@ describe("createFetchTracker long-lived stream watching (#2318)", () => {
     const updates: unknown[] = [];
     const fetcher = createFetchTracker((async () => response) as typeof fetch, {
       updateStream: (id, state) => updates.push({ id, state }),
+      streamUpdateIntervalMs: 0,
     });
     await expect(
       fetcher("https://example.com/mcp", { method: "GET" }),
@@ -971,6 +1057,7 @@ describe("createFetchTracker long-lived stream watching (#2318)", () => {
     const updates: unknown[] = [];
     const fetcher = createFetchTracker((async () => response) as typeof fetch, {
       updateStream: (id, state) => updates.push({ id, state }),
+      streamUpdateIntervalMs: 0,
     });
     await fetcher("https://example.com/mcp", { method: "GET" });
     await flush();

clients/web/src/test/core/mcp/inspectorClient-diagnostics-callbacks.test.ts

@@ -23,6 +23,12 @@ describe("InspectorClient diagnostics tracking callbacks", () => {
       // Never connects; the callbacks are exercised directly.
       { environment: { transport: () => ({}) as never } },
     );
+    // `createMessageTrackingCallbacks` is private: it is the seam the client
+    // hands its transport, and driving it directly is what lets this test
+    // exercise the bookkeeping without a connection. The double cast is the
+    // documented test-only exception for reaching a private member — the
+    // same one `inspectorClient-raw-wire.test.ts` uses — and `Internals`
+    // names exactly the one method it reaches.
     const callbacks = (
       client as unknown as Internals
     ).createMessageTrackingCallbacks();

clients/web/src/test/core/mcp/messageTrackingTransport.test.ts

@@ -48,18 +48,38 @@ describe("MessageTrackingTransport.send", () => {
     await expect(failing.send(request)).rejects.toBe(boom);
     expect(callbacks.trackRequest).toHaveBeenCalledWith(request, "client");
     expect(trackSendFailure).toHaveBeenCalledWith(request, boom);
-    // A caller's own abort is a cancellation, not a send failure: the
+    // A signal already aborted before the send began: the frame never went
+    // out, so this is a failed send and rolls back.
+    trackSendFailure.mockClear();
+    const preAborted = new AbortController();
+    preAborted.abort();
+    const preAbortedRequest = {
+      jsonrpc: "2.0",
+      id: 2,
+      method: "tools/call",
+    } as const;
+    await expect(
+      failing.send(preAbortedRequest, { requestSignal: preAborted.signal }),
+    ).rejects.toBe(boom);
+    expect(trackSendFailure).toHaveBeenCalledWith(preAbortedRequest, boom);
+    // A signal aborted while the send is in flight is a cancellation: the
     // request may well have reached the server, and stays unanswered.
     trackSendFailure.mockClear();
-    const controller = new AbortController();
-    controller.abort();
+    const midSend = new AbortController();
+    base.send = async () => {
+      midSend.abort();
+      throw boom;
+    };
     await expect(
       failing.send(
-        { jsonrpc: "2.0", id: 2, method: "tools/call" },
-        { requestSignal: controller.signal },
+        { jsonrpc: "2.0", id: 3, method: "tools/call" },
+        { requestSignal: midSend.signal },
       ),
     ).rejects.toBe(boom);
     expect(trackSendFailure).not.toHaveBeenCalled();
+    base.send = async () => {
+      throw boom;
+    };
     // Only requests roll back: a failed notification has no pending entry.
     trackSendFailure.mockClear();
     await expect(

core/mcp/fetchTracking.ts

@@ -250,8 +250,21 @@ export interface FetchTrackingCallbacks {
    * server serializes requests behind it (#2318).
    */
   updateStream?: (id: string, stream: FetchStreamState) => void;
+  /**
+   * How often, at most, `updateStream` reports a moving event count, in ms.
+   * The first report (open, zero events) and the close are immediate; counts
+   * in between are coalesced so a server pushing hundreds of notifications a
+   * second costs one update per window rather than one per event — each
+   * update re-publishes the whole fetch log and, on the web client, crosses
+   * the backend-to-browser event stream. Defaults to
+   * `DEFAULT_STREAM_UPDATE_INTERVAL_MS`; `0` reports every event, for tests.
+   */
+  streamUpdateIntervalMs?: number;
 }
 
+/** See `FetchTrackingCallbacks.streamUpdateIntervalMs`. */
+export const DEFAULT_STREAM_UPDATE_INTERVAL_MS = 250;
+
 /**
  * The framing a long-lived stream's events arrive in — the two content types
  * `isLongLivedStreamResponse` accepts. SSE dispatches on a blank line;
@@ -297,8 +310,10 @@ function watchLongLivedStream(
   framing: LongLivedStreamFraming,
   callbacks: FetchTrackingCallbacks,
 ): void {
-  const update = callbacks.updateStream;
-  if (!update) return;
+  const report = callbacks.updateStream;
+  if (!report) return;
+  const intervalMs =
+    callbacks.streamUpdateIntervalMs ?? DEFAULT_STREAM_UPDATE_INTERVAL_MS;
   let body: ReadableStream<Uint8Array> | null;
   try {
     body = response.clone().body;
@@ -313,23 +328,37 @@ function watchLongLivedStream(
   let eventCount = 0;
   let pending = "";
   let blockHasData = false;
+  // A moving count is reported at most once per interval (see
+  // `streamUpdateIntervalMs`); the open and the close go out at once.
+  let reportTimer: ReturnType<typeof setTimeout> | null = null;
+  const reportCount = (): void => {
+    if (intervalMs <= 0) {
+      report(id, { eventCount });
+      return;
+    }
+    if (reportTimer !== null) return;
+    reportTimer = setTimeout(() => {
+      reportTimer = null;
+      report(id, { eventCount });
+    }, intervalMs);
+  };
   // Announce the stream as open before a byte arrives. A stream that never
   // delivers anything is the #2187 tell, and it would otherwise be the one
   // stream with no state to show — the count only moves on an event.
-  update(id, { eventCount });
+  report(id, { eventCount });
 
   const consumeLine = (line: string): void => {
     if (framing === "ndjson") {
       if (line.trim() === "") return;
       eventCount += 1;
-      update(id, { eventCount });
+      reportCount();
       return;
     }
     if (line === "") {
       if (blockHasData) {
         eventCount += 1;
         blockHasData = false;
-        update(id, { eventCount });
+        reportCount();
       }
       return;
     }
@@ -368,17 +397,30 @@ function watchLongLivedStream(
         pending += decoder.decode(value, { stream: true });
         consumeCompleteLines();
       }
+      // A CR held back as a possible half of a CRLF is, at end-of-stream, a
+      // whole line ending — `data: x\r\r` is a complete event. Anything after
+      // it is a partial block and is dropped, as above.
+      if (pending.endsWith("\r")) consumeLine(pending.slice(0, -1));
     } catch {
       // An errored or aborted stream has still ended — fall through to report
       // the close; the entry's `error` field is for a fetch that never got a
       // response, which this one did.
     }
-    update(id, { eventCount, closedAt: new Date() });
+    if (reportTimer !== null) {
+      clearTimeout(reportTimer);
+      reportTimer = null;
+    }
+    try {
+      report(id, { eventCount, closedAt: new Date() });
+    } catch {
+      // A listener that throws must not become an unhandled rejection off a
+      // promise nothing awaits; the stream is closed either way.
+    }
   };
   // Deliberately not awaited: the fetch wrapper has to return the response
   // now, and this watcher lives as long as the stream does. It owns its own
-  // failures — the read loop is wrapped in a `try` that treats an error as a
-  // close — so the promise cannot reject.
+  // failures — the read loop and the final report are each wrapped in a
+  // `try` — so the promise cannot reject.
   void pump();
 }
 

core/mcp/messageTrackingTransport.ts

@@ -30,10 +30,11 @@ export interface MessageTrackingCallbacks {
    * `trackRequest` already recorded it, so a consumer keeping "requests still
    * awaiting a response" can roll that entry back — a request that never went
    * out is not unanswered (#2318). Not fired when the caller aborted the
-   * request through `requestSignal`: the modern transport aborts an in-flight
-   * request's own stream to cancel it (and the SDK does the same on a
-   * per-request timeout), and a request that was sent and then given up on
-   * is exactly the unanswered kind.
+   * request through `requestSignal` *during* the send: the modern transport
+   * aborts an in-flight request's own stream to cancel it (and the SDK does
+   * the same on a per-request timeout), and a request that was sent and then
+   * given up on is exactly the unanswered kind. A signal already aborted
+   * before the send began is a failed send like any other.
    */
   trackSendFailure?: (message: JSONRPCRequest, error: unknown) => void;
 }
@@ -109,8 +110,13 @@ export class MessageTrackingTransport implements Transport {
       } else if ("method" in message) {
         const request = message as JSONRPCRequest;
         this.callbacks.trackRequest?.(request, "client");
+        // A signal already aborted here means the frame will not go out at
+        // all — that is a failed send. One that becomes aborted while the
+        // send is in flight is a cancellation of a request that may well
+        // have reached the server, which stays unanswered.
+        const abortedBeforeSend = options?.requestSignal?.aborted === true;
         return this.baseTransport.send(message, options).catch((err) => {
-          if (!options?.requestSignal?.aborted) {
+          if (abortedBeforeSend || !options?.requestSignal?.aborted) {
             this.callbacks.trackSendFailure?.(request, err);
           }
           throw err;