fix(ui): hide synthetic transcript-repair history messages (#65458)

CHANGELOG.md

@@ -21,6 +21,7 @@ Docs: https://docs.openclaw.ai
 - Plugins/memory-core dreaming: keep bundled `memory-core` loaded alongside an explicit external memory slot owner only when that owner enables dreaming, while preserving `plugins.slots.memory = "none"` disable semantics. (#65411) Thanks @pradeep7127 and @vincentkoc.
 - Agents/Anthropic replay: preserve immutable signed-thinking replay safety across stored and live reruns, keep non-thinking embedded `tool_result` user blocks intact, and drop conflicting preserved tool IDs before validation so retries stop degrading into omitted tool calls. (#65126) Thanks @shakkernerd.
 - Telegram/direct sessions: keep commentary-only assistant fallback payloads out of visible direct delivery, so Codex planning chatter cannot leak into Telegram DMs when a run has no `final_answer` text. (#65112) Thanks @vincentkoc.
+- UI/WebChat: hide synthetic transcript-repair tool results from chat history reloads so internal recovery markers do not leak into visible chat after reconnects. (#65247) Thanks @wangwllu and @vincentkoc.
 - Infra/net: fix multipart FormData fields (including `model`) being silently dropped when a guarded runtime fetch body crosses a FormData implementation boundary, restoring OpenAI audio transcription requests that failed with HTTP 400. (#64349) Thanks @petr-sloup.
 - Dreaming/diary: use the host local timezone for diary timestamps when `dreaming.timezone` is unset, so `DREAMS.md` and the UI stop defaulting to UTC. (#65034) Thanks @neo1027144-creator and @vincentkoc.
 - Dreaming/diary: include the timezone abbreviation in diary timestamps so `DREAMS.md` and the UI make UTC or local host time explicit. (#65057) Thanks @Yanhu007 and @vincentkoc.

ui/src/ui/controllers/chat.test.ts

@@ -561,6 +561,66 @@ describe("loadChatHistory", () => {
     // text takes precedence — "real reply" is NOT silent, so message is kept.
     expect(state.chatMessages).toHaveLength(1);
   });
+
+  it("filters the synthetic transcript-repair tool result from history", async () => {
+    const messages = [
+      { role: "user", content: [{ type: "text", text: "hello" }] },
+      {
+        role: "toolResult",
+        toolCallId: "call_1",
+        toolName: "unknown",
+        isError: true,
+        content: [
+          {
+            type: "text",
+            text: "[openclaw] missing tool result in session history; inserted synthetic error result for transcript repair.",
+          },
+        ],
+      },
+      {
+        role: "toolResult",
+        toolCallId: "call_2",
+        toolName: "shell",
+        content: [{ type: "text", text: "real tool output" }],
+      },
+    ];
+    const mockClient = {
+      request: vi.fn().mockResolvedValue({ messages }),
+    };
+    const state = createState({
+      client: mockClient as unknown as ChatState["client"],
+      connected: true,
+    });
+
+    await loadChatHistory(state);
+
+    expect(state.chatMessages).toEqual([messages[0], messages[2]]);
+  });
+
+  it("keeps a user message even if it matches the synthetic repair text", async () => {
+    const messages = [
+      {
+        role: "user",
+        content: [
+          {
+            type: "text",
+            text: "[openclaw] missing tool result in session history; inserted synthetic error result for transcript repair.",
+          },
+        ],
+      },
+    ];
+    const mockClient = {
+      request: vi.fn().mockResolvedValue({ messages }),
+    };
+    const state = createState({
+      client: mockClient as unknown as ChatState["client"],
+      connected: true,
+    });
+
+    await loadChatHistory(state);
+
+    expect(state.chatMessages).toEqual(messages);
+  });
 });
 
 describe("sendChatMessage", () => {

ui/src/ui/controllers/chat.ts

@@ -11,6 +11,8 @@ import {
 } from "./scope-errors.ts";
 
 const SILENT_REPLY_PATTERN = /^\s*NO_REPLY\s*$/;
+const SYNTHETIC_TRANSCRIPT_REPAIR_RESULT =
+  "[openclaw] missing tool result in session history; inserted synthetic error result for transcript repair.";
 const chatHistoryRequestVersions = new WeakMap<object, number>();
 
 function beginChatHistoryRequest(state: ChatState): number {
@@ -53,6 +55,23 @@ function isAssistantSilentReply(message: unknown): boolean {
   return typeof text === "string" && isSilentReplyStream(text);
 }
 
+function isSyntheticTranscriptRepairToolResult(message: unknown): boolean {
+  if (!message || typeof message !== "object") {
+    return false;
+  }
+  const entry = message as Record<string, unknown>;
+  const role = normalizeLowercaseStringOrEmpty(entry.role);
+  if (role !== "toolresult") {
+    return false;
+  }
+  const text = extractText(message);
+  return typeof text === "string" && text.trim() === SYNTHETIC_TRANSCRIPT_REPAIR_RESULT;
+}
+
+function shouldHideHistoryMessage(message: unknown): boolean {
+  return isAssistantSilentReply(message) || isSyntheticTranscriptRepairToolResult(message);
+}
+
 export type ChatState = {
   client: GatewayBrowserClient | null;
   connected: boolean;
@@ -109,7 +128,7 @@ export async function loadChatHistory(state: ChatState) {
       return;
     }
     const messages = Array.isArray(res.messages) ? res.messages : [];
-    state.chatMessages = messages.filter((message) => !isAssistantSilentReply(message));
+    state.chatMessages = messages.filter((message) => !shouldHideHistoryMessage(message));
     state.chatThinkingLevel = res.thinkingLevel ?? null;
     // Clear all streaming state — history includes tool results and text
     // inline, so keeping streaming artifacts would cause duplicates.