fix(gateway): defer heartbeats during active replies

* fix(gateway): defer heartbeats during active replies

* fix(gateway): bind heartbeat reply run fallback

CHANGELOG.md

@@ -37,6 +37,7 @@ Docs: https://docs.openclaw.ai
 - CLI/setup: order the model/auth provider picker as OpenAI, Anthropic, xAI, Google, then the remaining providers alphabetically.
 - Gateway/diagnostics: add opt-in critical memory pressure stability snapshots with gateway logs, V8 heap, cgroup, active-resource, and redacted large session-file evidence. Fixes #82518.
 - Doctor/Gateway: avoid treating unrelated macOS LaunchAgents as legacy gateways just because their environment values mention old checkout paths.
+- Gateway/heartbeat: defer heartbeat runs while the target reply operation is queued or active, preventing heartbeat prompts from interleaving with WebChat responses before the streaming lane starts. Fixes #82722. Thanks @Andy-Xie-1145.
 - CLI/setup: collapse raw gateway config keys in existing-config summaries into friendly `Model` and `Gateway` rows.
 - CLI/config: show concise human config-write output with an indented backup path instead of printing checksum-heavy overwrite audit details by default.
 - CLI/docs: call the canonical lowercase docs MCP search tool and surface MCP errors instead of returning empty search results. Fixes #82702. (#82704) Thanks @hclsys.

src/infra/heartbeat-runner.skips-busy-session-lane.test.ts

@@ -228,6 +228,28 @@ describe("heartbeat runner skips when target session lane is busy", () => {
     });
   });
 
+  it("returns requests-in-flight when the target session has an active reply run", async () => {
+    await withTempHeartbeatSandbox(async ({ storePath, replySpy }) => {
+      const cfg = createHeartbeatTelegramConfig();
+      const sessionKey = await seedHeartbeatTelegramSession(storePath, cfg);
+      const isReplyRunActive = vi.fn((key: string) => key === sessionKey);
+
+      const result = await runHeartbeatOnce({
+        cfg,
+        deps: {
+          getQueueSize: vi.fn((_lane?: string) => 0),
+          isReplyRunActive,
+          nowMs: () => Date.now(),
+          getReplyFromConfig: replySpy,
+        } as HeartbeatDeps,
+      });
+
+      expect(result).toEqual({ status: "skipped", reason: HEARTBEAT_SKIP_REQUESTS_IN_FLIGHT });
+      expect(isReplyRunActive).toHaveBeenCalledWith(sessionKey);
+      expect(replySpy).not.toHaveBeenCalled();
+    });
+  });
+
   it("does not defer on a recent heartbeat ack pending final delivery", async () => {
     await withTempHeartbeatSandbox(async ({ storePath, replySpy }) => {
       const cfg = createHeartbeatTelegramConfig();

src/infra/heartbeat-runner.ts

@@ -34,6 +34,7 @@ import {
   type HeartbeatTask,
 } from "../auto-reply/heartbeat.js";
 import { resolveDefaultModel } from "../auto-reply/reply/directive-handling.defaults.js";
+import { replyRunRegistry } from "../auto-reply/reply/reply-run-registry.js";
 import { resolveResponsePrefixTemplate } from "../auto-reply/reply/response-prefix-template.js";
 import { HEARTBEAT_TOKEN } from "../auto-reply/tokens.js";
 import type { ReplyPayload } from "../auto-reply/types.js";
@@ -142,6 +143,7 @@ export type HeartbeatDeps = OutboundSendDeps &
     runtime?: RuntimeEnv;
     getQueueSize?: (lane?: string) => number;
     getCommandLaneSnapshots?: () => readonly CommandLaneSnapshot[];
+    isReplyRunActive?: (sessionKey: string) => boolean;
     nowMs?: () => number;
   };
 
@@ -1350,6 +1352,16 @@ export async function runHeartbeatOnce(opts: {
     return { status: "skipped", reason: preflight.skipReason };
   }
   const { entry, sessionKey, storePath, suppressOriginatingContext } = preflight.session;
+  const isReplyRunActive =
+    opts.deps?.isReplyRunActive ?? ((key: string) => replyRunRegistry.isActive(key));
+  if (isReplyRunActive(sessionKey)) {
+    emitHeartbeatEvent({
+      status: "skipped",
+      reason: HEARTBEAT_SKIP_REQUESTS_IN_FLIGHT,
+      durationMs: Date.now() - startedAt,
+    });
+    return { status: "skipped", reason: HEARTBEAT_SKIP_REQUESTS_IN_FLIGHT };
+  }
 
   // Check the resolved session lane — if it is busy, skip to avoid interrupting
   // an active streaming turn.  The wake-layer retry (heartbeat-wake.ts) will