fix: speed up Telegram status diagnostics

extensions/telegram/src/bot-message.test.ts

@@ -76,7 +76,10 @@ describe("telegram bot message processor", () => {
     sendMessage: ReturnType<typeof vi.fn>,
   ) {
     const runtimeError = vi.fn();
-    buildTelegramMessageContext.mockResolvedValue(context);
+    buildTelegramMessageContext.mockResolvedValue({
+      sendTyping: vi.fn().mockResolvedValue(undefined),
+      ...context,
+    });
     dispatchTelegramMessage.mockRejectedValue(new Error("dispatch exploded"));
     const processMessage = createTelegramMessageProcessor({
       ...baseDeps,
@@ -87,12 +90,21 @@ describe("telegram bot message processor", () => {
   }
 
   it("dispatches when context is available", async () => {
-    buildTelegramMessageContext.mockResolvedValue({ route: { sessionKey: "agent:main:main" } });
+    const sendTyping = vi.fn().mockResolvedValue(undefined);
+    buildTelegramMessageContext.mockResolvedValue({
+      chatId: 123,
+      route: { sessionKey: "agent:main:main" },
+      sendTyping,
+    });
 
     const processMessage = createTelegramMessageProcessor(baseDeps);
     await processSampleMessage(processMessage);
 
+    expect(sendTyping).toHaveBeenCalledTimes(1);
     expect(dispatchTelegramMessage).toHaveBeenCalledTimes(1);
+    expect(sendTyping.mock.invocationCallOrder[0]).toBeLessThan(
+      dispatchTelegramMessage.mock.invocationCallOrder[0],
+    );
   });
 
   it("skips dispatch when no context is produced", async () => {
@@ -102,6 +114,21 @@ describe("telegram bot message processor", () => {
     expect(dispatchTelegramMessage).not.toHaveBeenCalled();
   });
 
+  it("keeps dispatch running when the early typing cue fails", async () => {
+    const sendTyping = vi.fn().mockRejectedValue(new Error("typing failed"));
+    buildTelegramMessageContext.mockResolvedValue({
+      chatId: 123,
+      route: { sessionKey: "agent:main:main" },
+      sendTyping,
+    });
+
+    const processMessage = createTelegramMessageProcessor(baseDeps);
+    await processSampleMessage(processMessage);
+
+    expect(sendTyping).toHaveBeenCalledTimes(1);
+    expect(dispatchTelegramMessage).toHaveBeenCalledTimes(1);
+  });
+
   it("sends user-visible fallback when dispatch throws", async () => {
     const sendMessage = vi.fn().mockResolvedValue(undefined);
     const { processMessage, runtimeError } = createDispatchFailureHarness(

extensions/telegram/src/bot-message.ts

@@ -107,6 +107,9 @@ export const createTelegramMessageProcessor = (deps: TelegramMessageProcessorDep
           (options?.ingressBuffer ? ` buffer=${options.ingressBuffer}` : ""),
       );
     }
+    void context.sendTyping().catch((err) => {
+      logVerbose(`telegram early typing cue failed for chat ${context.chatId}: ${String(err)}`);
+    });
     try {
       await dispatchTelegramMessage({
         context,