fix(telegram): sanitize invalid stream-order errors (#55999)

* fix(telegram): sanitize invalid stream order errors

* docs(changelog): remove maintainer follow-up entry

CHANGELOG.md

@@ -115,6 +115,7 @@ Docs: https://docs.openclaw.ai
 - Message tool/buttons: keep the shared `buttons` schema optional in merged tool definitions so plain `action=send` calls stop failing validation when no buttons are provided. (#54418) Thanks @adzendo.
 - Agents/openai-compatible tool calls: deduplicate repeated tool call ids across live assistant messages and replayed history so OpenAI-compatible backends no longer reject duplicate `tool_call_id` values with HTTP 400. (#40996) Thanks @xaeon2026.
 - Models/openai-completions: default non-native OpenAI-compatible providers to omit tool-definition `strict` fields unless users explicitly opt back in, so tool calling keeps working on providers that reject that option. (#45497) Thanks @sahancava.
+- Telegram/Anthropic streaming: replace raw invalid stream-order provider errors with a safe retry message so internal `message_start/message_stop` failures do not leak into chats. (#55408) Thanks @imydal.
 - Plugins/context engines: retry strict legacy `assemble()` calls without the new `prompt` field when older engines reject it, preserving prompt-aware retrieval compatibility for pre-prompt plugins. (#50848) thanks @danhdoan.
 - CLI/update status: explicitly say `up to date` when the local version already matches npm latest, while keeping the availability logic unchanged. (#51409) Thanks @dongzhenye.
 - Daemon/Linux: stop flagging non-gateway systemd services as duplicate gateways just because their unit files mention OpenClaw, reducing false-positive doctor/log noise. (#45328) Thanks @gregretkowski.

src/agents/pi-embedded-helpers.formatassistanterrortext.test.ts

@@ -204,6 +204,15 @@ describe("formatAssistantErrorText", () => {
       "LLM request failed: network connection was interrupted.",
     );
   });
+
+  it("sanitizes invalid streaming event order errors", () => {
+    const msg = makeAssistantError(
+      'Unexpected event order, got message_start before receiving "message_stop"',
+    );
+    expect(formatAssistantErrorText(msg)).toBe(
+      "LLM request failed: provider returned an invalid streaming response. Please try again.",
+    );
+  });
 });
 
 describe("formatRawAssistantErrorForUi", () => {

src/agents/pi-embedded-helpers.sanitizeuserfacingtext.test.ts

@@ -136,6 +136,17 @@ describe("sanitizeUserFacingText", () => {
     ).toBe("LLM request failed: connection refused by the provider endpoint.");
   });
 
+  it("sanitizes invalid streaming event order errors", () => {
+    expect(
+      sanitizeUserFacingText(
+        'Unexpected event order, got message_start before receiving "message_stop"',
+        { errorContext: true },
+      ),
+    ).toBe(
+      "LLM request failed: provider returned an invalid streaming response. Please try again.",
+    );
+  });
+
   it.each([
     {
       input: "Hello there!\n\nHello there!",

src/agents/pi-embedded-helpers/errors.ts

@@ -174,6 +174,18 @@ function isReasoningConstraintErrorMessage(raw: string): boolean {
   );
 }
 
+function isInvalidStreamingEventOrderError(raw: string): boolean {
+  if (!raw) {
+    return false;
+  }
+  const lower = raw.toLowerCase();
+  return (
+    lower.includes("unexpected event order") &&
+    lower.includes("message_start") &&
+    lower.includes("message_stop")
+  );
+}
+
 function hasRateLimitTpmHint(raw: string): boolean {
   const lower = raw.toLowerCase();
   return /\btpm\b/i.test(lower) || lower.includes("tokens per minute");
@@ -685,6 +697,10 @@ export function formatAssistantErrorText(
     );
   }
 
+  if (isInvalidStreamingEventOrderError(raw)) {
+    return "LLM request failed: provider returned an invalid streaming response. Please try again.";
+  }
+
   // Catch role ordering errors - including JSON-wrapped and "400" prefix variants
   if (
     /incorrect role information|roles must alternate|400.*role|"message".*role.*information/i.test(
@@ -777,6 +793,10 @@ export function sanitizeUserFacingText(text: string, opts?: { errorContext?: boo
       return BILLING_ERROR_USER_MESSAGE;
     }
 
+    if (isInvalidStreamingEventOrderError(trimmed)) {
+      return "LLM request failed: provider returned an invalid streaming response. Please try again.";
+    }
+
     if (isRawApiErrorPayload(trimmed) || isLikelyHttpErrorText(trimmed)) {
       return formatRawAssistantErrorForUi(trimmed);
     }