fix: track diagnostic progress before stuck warnings

CHANGELOG.md

@@ -25,6 +25,7 @@ Docs: https://docs.openclaw.ai
 - Channels/status reactions: remove stale non-terminal lifecycle reactions when a run reaches done or error, so Discord does not leave a permanent thinking emoji after completion. Fixes #75458. Thanks @davelutztx.
 - Discord/doctor: migrate unsupported per-channel `agentId` entries under guild channel config into top-level `bindings[]` routes, so `openclaw doctor --fix` preserves the intended agent route instead of stripping it as an unknown key. Fixes #62455. Thanks @lobster-biscuit.
 - Gateway/config: log config health-state write failures instead of silently hiding config observe-recovery write errors. Thanks @sallyom.
+- Diagnostics: reset stuck-session timers on reply, tool, status, block, and ACP progress events, and back off repeated `session.stuck` diagnostics while a session remains unchanged. Supersedes #72010. Thanks @rubencu.
 
 ## 2026.4.30
 

docs/.generated/config-baseline.sha256

@@ -1,4 +1,4 @@
-74530fefef9ed55cab302802bc0be413ec56929e73c12d4bf4f1e4d290813adc  config-baseline.json
+8bbb620e445cba64aa8a451cfc1a7142ac24e8c80088d74a2fc813ee9e221680  config-baseline.json
-21db87c2ebec8844e20bf66ea474c08f3adab842234ff334870fe3e8d87995b4  config-baseline.core.json
+d145a87759d16d5f58873db337a25cb134ab25e776cd454812dca99bb9cb12a7  config-baseline.core.json
 c401cd3450f1737bc92418cfea301d20b54b7fbef9e6049834acc01af338e538  config-baseline.channel.json
 7731a0b93cb335b56fac4c807447ba659fea51ea7a6cd844dc0ef5616669ee75  config-baseline.plugin.json

docs/concepts/agent-loop.md

@@ -164,7 +164,7 @@ surfaces, while Codex native hooks remain a separate lower-level Codex mechanism
 - `agent.wait` default: 30s (just the wait). `timeoutMs` param overrides.
 - Agent runtime: `agents.defaults.timeoutSeconds` default 172800s (48 hours); enforced in `runEmbeddedPiAgent` abort timer.
 - Cron runtime: isolated agent-turn `timeoutSeconds` is owned by cron. The scheduler starts that timer when execution begins, aborts the underlying run at the configured deadline, then runs bounded cleanup before recording the timeout so a stale child session cannot keep the lane stuck.
-- Session liveness diagnostics: with diagnostics enabled, `diagnostics.stuckSessionWarnMs` classifies long `processing` sessions. Active embedded runs, model calls, and tool calls report as `session.long_running`; active work with no recent progress reports as `session.stalled`; `session.stuck` is reserved for stale session bookkeeping with no active work, and only that path releases the affected session lane so queued startup work can drain.
+- Session liveness diagnostics: with diagnostics enabled, `diagnostics.stuckSessionWarnMs` classifies long `processing` sessions that have no observed reply, tool, status, block, or ACP progress. Active embedded runs, model calls, and tool calls report as `session.long_running`; active work with no recent progress reports as `session.stalled`; `session.stuck` is reserved for stale session bookkeeping with no active work, and only that path releases the affected session lane so queued startup work can drain. Repeated `session.stuck` diagnostics back off while the session remains unchanged.
 - Model idle timeout: OpenClaw aborts a model request when no response chunks arrive before the idle window. `models.providers.<id>.timeoutSeconds` extends this idle watchdog for slow local/self-hosted providers; otherwise OpenClaw uses `agents.defaults.timeoutSeconds` when configured, capped at 120s by default. Cron-triggered runs with no explicit model or agent timeout disable the idle watchdog and rely on the cron outer timeout.
 - Provider HTTP request timeout: `models.providers.<id>.timeoutSeconds` applies to that provider's model HTTP fetches, including connect, headers, body, SDK request timeout, total guarded-fetch abort handling, and model stream idle watchdog. Use this for slow local/self-hosted providers such as Ollama before raising the whole agent runtime timeout.
 

docs/concepts/queue.md

@@ -115,7 +115,7 @@ keys.
 - If commands seem stuck, enable verbose logs and look for “queued for …ms” lines to confirm the queue is draining.
 - If you need queue depth, enable verbose logs and watch for queue timing lines.
 - Codex app-server runs that accept a turn and then stop emitting progress are interrupted by the Codex adapter so the active session lane can release instead of waiting for the outer run timeout.
-- When diagnostics are enabled, sessions that remain in `processing` past `diagnostics.stuckSessionWarnMs` are classified by current activity. Active work logs as `session.long_running`; active work with no recent progress logs as `session.stalled`; `session.stuck` is reserved for stale session bookkeeping with no active work, and only that path can release the affected session lane so queued work drains.
+- When diagnostics are enabled, sessions that remain in `processing` past `diagnostics.stuckSessionWarnMs` with no observed reply, tool, status, block, or ACP progress are classified by current activity. Active work logs as `session.long_running`; active work with no recent progress logs as `session.stalled`; `session.stuck` is reserved for stale session bookkeeping with no active work, and only that path can release the affected session lane so queued work drains. Repeated `session.stuck` diagnostics back off while the session remains unchanged.
 
 ## Related
 

docs/gateway/configuration-reference.md

@@ -937,7 +937,7 @@ Notes:
 
 - `enabled`: master toggle for instrumentation output (default: `true`).
 - `flags`: array of flag strings enabling targeted log output (supports wildcards like `"telegram.*"` or `"*"`).
-- `stuckSessionWarnMs`: age threshold in ms for classifying long-running processing sessions as `session.long_running`, `session.stalled`, or `session.stuck`.
+- `stuckSessionWarnMs`: no-progress age threshold in ms for classifying long-running processing sessions as `session.long_running`, `session.stalled`, or `session.stuck`. Reply, tool, status, block, and ACP progress reset the timer; repeated `session.stuck` diagnostics back off while unchanged.
 - `otel.enabled`: enables the OpenTelemetry export pipeline (default: `false`). For the full configuration, signal catalog, and privacy model, see [OpenTelemetry export](/gateway/opentelemetry).
 - `otel.endpoint`: collector URL for OTel export.
 - `otel.tracesEndpoint` / `otel.metricsEndpoint` / `otel.logsEndpoint`: optional signal-specific OTLP endpoints. When set, they override `otel.endpoint` for that signal only.

src/auto-reply/reply/acp-projector.test.ts

@@ -5,7 +5,10 @@ import { createAcpTestConfig as createCfg } from "./test-fixtures/acp-runtime.js
 
 type Delivery = { kind: string; text?: string };
 
-function createProjectorHarness(cfgOverrides?: Parameters<typeof createCfg>[0]) {
+function createProjectorHarness(
+  cfgOverrides?: Parameters<typeof createCfg>[0],
+  opts?: { onProgress?: () => void },
+) {
   const deliveries: Delivery[] = [];
   const projector = createAcpReplyProjector({
     cfg: createCfg(cfgOverrides),
@@ -14,6 +17,7 @@ function createProjectorHarness(cfgOverrides?: Parameters<typeof createCfg>[0])
       deliveries.push({ kind, text: payload.text });
       return true;
     },
+    onProgress: opts?.onProgress,
   });
   return { deliveries, projector };
 }
@@ -175,6 +179,28 @@ async function runHiddenBoundaryCase(params: {
 }
 
 describe("createAcpReplyProjector", () => {
+  it("reports progress for ACP runtime events before delivery filtering", async () => {
+    const onProgress = vi.fn();
+    const { projector } = createProjectorHarness(undefined, { onProgress });
+
+    await projector.onEvent({
+      type: "text_delta",
+      stream: "thought",
+      text: "hidden reasoning",
+      tag: "agent_message_chunk",
+    });
+    await projector.onEvent({
+      type: "tool_call",
+      tag: "tool_call",
+      toolCallId: "tool-1",
+      status: "in_progress",
+      title: "Run command",
+      text: "Run command",
+    });
+
+    expect(onProgress).toHaveBeenCalledTimes(2);
+  });
+
   it("coalesces text deltas into bounded block chunks", async () => {
     const { deliveries, projector } = createProjectorHarness();
 

src/auto-reply/reply/acp-projector.ts

@@ -173,6 +173,7 @@ export function createAcpReplyProjector(params: {
     payload: ReplyPayload,
     meta?: AcpProjectedDeliveryMeta,
   ) => Promise<boolean>;
+  onProgress?: () => void;
   provider?: string;
   accountId?: string;
 }): AcpReplyProjector {
@@ -403,6 +404,7 @@ export function createAcpReplyProjector(params: {
   };
 
   const onEvent = async (event: AcpRuntimeEvent): Promise<void> => {
+    params.onProgress?.();
     if (event.type === "text_delta") {
       if (event.stream && event.stream !== "output") {
         return;

src/auto-reply/reply/dispatch-acp.test.ts

@@ -74,6 +74,10 @@ const mediaUnderstandingMocks = vi.hoisted(() => ({
   applyMediaUnderstanding: vi.fn(async (_params: unknown) => undefined),
 }));
 
+const diagnosticMocks = vi.hoisted(() => ({
+  markDiagnosticSessionProgress: vi.fn(),
+}));
+
 const sessionMetaMocks = vi.hoisted(() => ({
   readAcpSessionEntry: vi.fn<
     (params: { sessionKey: string; cfg?: OpenClawConfig }) => AcpSessionStoreEntry | null
@@ -168,6 +172,10 @@ vi.mock("./dispatch-acp-session.runtime.js", () => ({
     sessionMetaMocks.readAcpSessionEntry(params),
 }));
 
+vi.mock("../../logging/diagnostic.js", () => ({
+  markDiagnosticSessionProgress: diagnosticMocks.markDiagnosticSessionProgress,
+}));
+
 vi.mock("./dispatch-acp-transcript.runtime.js", () => ({
   persistAcpDispatchTranscript: (params: unknown) =>
     transcriptMocks.persistAcpDispatchTranscript(params),
@@ -374,6 +382,7 @@ describe("tryDispatchAcpReply", () => {
     ttsMocks.resolveTtsConfig.mockReturnValue({ mode: "final" });
     mediaUnderstandingMocks.applyMediaUnderstanding.mockReset();
     mediaUnderstandingMocks.applyMediaUnderstanding.mockResolvedValue(undefined);
+    diagnosticMocks.markDiagnosticSessionProgress.mockReset();
     sessionMetaMocks.readAcpSessionEntry.mockReset();
     sessionMetaMocks.readAcpSessionEntry.mockReturnValue(null);
     transcriptMocks.persistAcpDispatchTranscript.mockClear();
@@ -545,6 +554,18 @@ describe("tryDispatchAcpReply", () => {
     expect(onReplyStart).toHaveBeenCalledTimes(1);
   });
 
+  it("does not mark ACP diagnostic progress when diagnostics are disabled", async () => {
+    setReadyAcpResolution();
+    mockVisibleTextTurn();
+
+    await runDispatch({
+      bodyForAgent: "visible",
+      cfg: createAcpTestConfig({ diagnostics: { enabled: false } }),
+    });
+
+    expect(diagnosticMocks.markDiagnosticSessionProgress).not.toHaveBeenCalled();
+  });
+
   it("does not start reply lifecycle for empty ACP prompt", async () => {
     setReadyAcpResolution();
     const onReplyStart = vi.fn();

src/auto-reply/reply/dispatch-acp.ts

@@ -11,9 +11,11 @@ import type { OpenClawConfig } from "../../config/types.openclaw.js";
 import type { TtsAutoMode } from "../../config/types.tts.js";
 import { logVerbose } from "../../globals.js";
 import { emitAgentEvent } from "../../infra/agent-events.js";
+import { isDiagnosticsEnabled } from "../../infra/diagnostic-events.js";
 import { formatErrorMessage } from "../../infra/errors.js";
 import { generateSecureUuid } from "../../infra/secure-random.js";
 import { prefixSystemMessage } from "../../infra/system-message.js";
+import { markDiagnosticSessionProgress } from "../../logging/diagnostic.js";
 import { resolveAgentIdFromSessionKey } from "../../routing/session-key.js";
 import {
   normalizeLowercaseStringOrEmpty,
@@ -342,6 +344,23 @@ export async function tryDispatchAcpReply(params: {
   }
   const canonicalSessionKey = acpResolution.sessionKey;
   const acpAgentId = resolveAgentIdFromSessionKey(canonicalSessionKey);
+  const progressSessionKeys = isDiagnosticsEnabled(params.cfg)
+    ? Array.from(
+        new Set(
+          [params.ctx.SessionKey, sessionKey, canonicalSessionKey]
+            .map((key) => normalizeOptionalString(key))
+            .filter((key): key is string => Boolean(key)),
+        ),
+      )
+    : [];
+  const markAcpProgress =
+    progressSessionKeys.length > 0
+      ? () => {
+          for (const key of progressSessionKeys) {
+            markDiagnosticSessionProgress({ sessionKey: key });
+          }
+        }
+      : undefined;
 
   let queuedFinal = false;
   const delivery = createAcpDispatchDeliveryCoordinator({
@@ -401,6 +420,7 @@ export async function tryDispatchAcpReply(params: {
     cfg: params.cfg,
     shouldSendToolSummaries: params.shouldSendToolSummaries,
     deliver: delivery.deliver,
+    onProgress: markAcpProgress,
     provider: params.ctx.Surface ?? params.ctx.Provider,
     accountId: effectiveDispatchAccountId,
   });

src/auto-reply/reply/dispatch-from-config.acp-abort.test.ts

@@ -178,6 +178,7 @@ describe("dispatchReplyFromConfig ACP abort", () => {
     diagnosticMocks.logMessageQueued.mockReset();
     diagnosticMocks.logMessageProcessed.mockReset();
     diagnosticMocks.logSessionStateChange.mockReset();
+    diagnosticMocks.markDiagnosticSessionProgress.mockReset();
     agentEventMocks.emitAgentEvent.mockReset();
     agentEventMocks.onAgentEvent.mockReset().mockImplementation(() => () => {});
     setNoAbort();

src/auto-reply/reply/dispatch-from-config.reply-dispatch.test.ts

@@ -84,6 +84,7 @@ describe("dispatchReplyFromConfig reply_dispatch hook", () => {
     diagnosticMocks.logMessageQueued.mockReset();
     diagnosticMocks.logMessageProcessed.mockReset();
     diagnosticMocks.logSessionStateChange.mockReset();
+    diagnosticMocks.markDiagnosticSessionProgress.mockReset();
     runtimePluginMocks.ensureRuntimePluginsLoaded.mockReset();
     resetPluginTtsAndThreadMocks();
   });

src/auto-reply/reply/dispatch-from-config.shared.test-harness.ts

@@ -28,6 +28,7 @@ const diagnosticMocks = vi.hoisted(() => ({
   logMessageQueued: vi.fn(),
   logMessageProcessed: vi.fn(),
   logSessionStateChange: vi.fn(),
+  markDiagnosticSessionProgress: vi.fn(),
 }));
 const hookMocks = vi.hoisted(() => ({
   registry: {
@@ -177,6 +178,7 @@ vi.mock("../../logging/diagnostic.js", () => ({
   logMessageQueued: diagnosticMocks.logMessageQueued,
   logMessageProcessed: diagnosticMocks.logMessageProcessed,
   logSessionStateChange: diagnosticMocks.logSessionStateChange,
+  markDiagnosticSessionProgress: diagnosticMocks.markDiagnosticSessionProgress,
 }));
 vi.mock("../../config/sessions/thread-info.js", () => ({
   parseSessionThreadInfo: (sessionKey: string | undefined) =>

src/auto-reply/reply/dispatch-from-config.test.ts

@@ -42,6 +42,7 @@ const diagnosticMocks = vi.hoisted(() => ({
   logMessageQueued: vi.fn(),
   logMessageProcessed: vi.fn(),
   logSessionStateChange: vi.fn(),
+  markDiagnosticSessionProgress: vi.fn(),
 }));
 const hookMocks = vi.hoisted(() => ({
   registry: {
@@ -343,6 +344,7 @@ vi.mock("../../logging/diagnostic.js", () => ({
   logMessageQueued: diagnosticMocks.logMessageQueued,
   logMessageProcessed: diagnosticMocks.logMessageProcessed,
   logSessionStateChange: diagnosticMocks.logSessionStateChange,
+  markDiagnosticSessionProgress: diagnosticMocks.markDiagnosticSessionProgress,
 }));
 vi.mock("../../config/sessions/thread-info.js", () => ({
   parseSessionThreadInfo: (sessionKey: string | undefined) =>
@@ -749,6 +751,7 @@ describe("dispatchReplyFromConfig", () => {
     diagnosticMocks.logMessageQueued.mockClear();
     diagnosticMocks.logMessageProcessed.mockClear();
     diagnosticMocks.logSessionStateChange.mockClear();
+    diagnosticMocks.markDiagnosticSessionProgress.mockClear();
     hookMocks.runner.hasHooks.mockClear();
     hookMocks.runner.hasHooks.mockImplementation(
       (hookName?: string) => hookName === "reply_dispatch",
@@ -2959,6 +2962,89 @@ describe("dispatchReplyFromConfig", () => {
     );
   });
 
+  it("marks diagnostic progress for real reply events but not reply start callbacks", async () => {
+    setNoAbort();
+    const cfg = { diagnostics: { enabled: true } } as OpenClawConfig;
+    const dispatcher = createDispatcher();
+    const ctx = buildTestCtx({
+      Provider: "slack",
+      Surface: "slack",
+      SessionKey: "agent:main:main",
+      To: "slack:C123",
+    });
+    const onReplyStart = vi.fn(async () => {});
+    const replyResolver = async (
+      _ctx: MsgContext,
+      opts?: GetReplyOptions,
+    ): Promise<ReplyPayload> => {
+      await opts?.onReplyStart?.();
+      await opts?.onToolResult?.({ text: "tool progress" });
+      return { text: "hi" };
+    };
+
+    await dispatchReplyFromConfig({
+      ctx,
+      cfg,
+      dispatcher,
+      replyOptions: { onReplyStart },
+      replyResolver,
+    });
+
+    expect(onReplyStart).toHaveBeenCalledTimes(1);
+    expect(diagnosticMocks.markDiagnosticSessionProgress).toHaveBeenCalledTimes(1);
+    expect(diagnosticMocks.markDiagnosticSessionProgress).toHaveBeenCalledWith({
+      sessionKey: "agent:main:main",
+    });
+  });
+
+  it("keeps diagnostic progress when source progress callbacks are suppressed", async () => {
+    setNoAbort();
+    const cfg = { diagnostics: { enabled: true } } as OpenClawConfig;
+    const dispatcher = createDispatcher();
+    const ctx = buildTestCtx({
+      Provider: "discord",
+      Surface: "discord",
+      ChatType: "channel",
+      SessionKey: "agent:main:discord:channel:C1",
+      To: "discord:channel:C1",
+    });
+    const callbacks = {
+      toolStart: vi.fn(async () => {}),
+      itemEvent: vi.fn(async () => {}),
+      commandOutput: vi.fn(async () => {}),
+    };
+    const replyResolver = async (
+      _ctx: MsgContext,
+      opts?: GetReplyOptions,
+    ): Promise<ReplyPayload> => {
+      await opts?.onToolStart?.({ name: "lookup" });
+      await opts?.onItemEvent?.({ progressText: "working" });
+      await opts?.onCommandOutput?.({ output: "line", status: "running" });
+      return { text: "hi" };
+    };
+
+    await dispatchReplyFromConfig({
+      ctx,
+      cfg,
+      dispatcher,
+      replyOptions: {
+        sourceReplyDeliveryMode: "message_tool_only",
+        onToolStart: callbacks.toolStart,
+        onItemEvent: callbacks.itemEvent,
+        onCommandOutput: callbacks.commandOutput,
+      },
+      replyResolver,
+    });
+
+    expect(callbacks.toolStart).not.toHaveBeenCalled();
+    expect(callbacks.itemEvent).not.toHaveBeenCalled();
+    expect(callbacks.commandOutput).not.toHaveBeenCalled();
+    expect(diagnosticMocks.markDiagnosticSessionProgress).toHaveBeenCalledTimes(3);
+    expect(diagnosticMocks.markDiagnosticSessionProgress).toHaveBeenCalledWith({
+      sessionKey: "agent:main:discord:channel:C1",
+    });
+  });
+
   it("routes plugin-owned bindings to the owning plugin before generic inbound claim broadcast", async () => {
     setNoAbort();
     hookMocks.runner.hasHooks.mockImplementation(

src/auto-reply/reply/dispatch-from-config.ts

@@ -45,6 +45,7 @@ import {
   logMessageProcessed,
   logMessageQueued,
   logSessionStateChange,
+  markDiagnosticSessionProgress,
 } from "../../logging/diagnostic.js";
 import {
   buildPluginBindingDeclinedText,
@@ -411,6 +412,15 @@ export async function dispatchReplyFromConfig(
   const boundAcpDispatchSessionKey = resolveBoundAcpDispatchSessionKey({ ctx, cfg });
   const acpDispatchSessionKey =
     boundAcpDispatchSessionKey ?? initialSessionStoreEntry.sessionKey ?? sessionKey;
+  const markProgress = () => {
+    if (!canTrackSession || !sessionKey) {
+      return;
+    }
+    markDiagnosticSessionProgress({ sessionKey });
+    if (acpDispatchSessionKey && acpDispatchSessionKey !== sessionKey) {
+      markDiagnosticSessionProgress({ sessionKey: acpDispatchSessionKey });
+    }
+  };
   const sessionStoreEntry = boundAcpDispatchSessionKey
     ? resolveSessionStoreLookup({ ...ctx, SessionKey: boundAcpDispatchSessionKey }, cfg)
     : initialSessionStoreEntry;
@@ -1190,6 +1200,19 @@ export async function dispatchReplyFromConfig(
     const onPlanUpdateFromReplyOptions = params.replyOptions?.onPlanUpdate;
     const onApprovalEventFromReplyOptions = params.replyOptions?.onApprovalEvent;
     const onPatchSummaryFromReplyOptions = params.replyOptions?.onPatchSummary;
+    const wrapProgressCallback = <Args extends unknown[]>(
+      callback: ((...args: Args) => Promise<void> | void) | undefined,
+    ): ((...args: Args) => Promise<void>) | undefined => {
+      if (!callback && (!suppressAutomaticSourceDelivery || !canTrackSession)) {
+        return undefined;
+      }
+      return async (...args: Args) => {
+        markProgress();
+        if (!suppressAutomaticSourceDelivery) {
+          await callback?.(...args);
+        }
+      };
+    };
 
     const replyResolver =
       params.replyResolver ?? (await loadGetReplyFromConfigRuntime()).getReplyFromConfig;
@@ -1203,33 +1226,18 @@ export async function dispatchReplyFromConfig(
         sourceReplyDeliveryMode,
         typingPolicy: typing.typingPolicy,
         suppressTyping: typing.suppressTyping,
-        onPartialReply: suppressAutomaticSourceDelivery
+        onPartialReply: wrapProgressCallback(params.replyOptions?.onPartialReply),
-          ? undefined
+        onReasoningStream: wrapProgressCallback(params.replyOptions?.onReasoningStream),
-          : params.replyOptions?.onPartialReply,
+        onReasoningEnd: wrapProgressCallback(params.replyOptions?.onReasoningEnd),
-        onReasoningStream: suppressAutomaticSourceDelivery
+        onAssistantMessageStart: wrapProgressCallback(params.replyOptions?.onAssistantMessageStart),
-          ? undefined
+        onBlockReplyQueued: wrapProgressCallback(params.replyOptions?.onBlockReplyQueued),
-          : params.replyOptions?.onReasoningStream,
+        onToolStart: wrapProgressCallback(params.replyOptions?.onToolStart),
-        onReasoningEnd: suppressAutomaticSourceDelivery
+        onItemEvent: wrapProgressCallback(params.replyOptions?.onItemEvent),
-          ? undefined
+        onCommandOutput: wrapProgressCallback(params.replyOptions?.onCommandOutput),
-          : params.replyOptions?.onReasoningEnd,
+        onCompactionStart: wrapProgressCallback(params.replyOptions?.onCompactionStart),
-        onAssistantMessageStart: suppressAutomaticSourceDelivery
+        onCompactionEnd: wrapProgressCallback(params.replyOptions?.onCompactionEnd),
-          ? undefined
-          : params.replyOptions?.onAssistantMessageStart,
-        onBlockReplyQueued: suppressAutomaticSourceDelivery
-          ? undefined
-          : params.replyOptions?.onBlockReplyQueued,
-        onToolStart: suppressAutomaticSourceDelivery ? undefined : params.replyOptions?.onToolStart,
-        onItemEvent: suppressAutomaticSourceDelivery ? undefined : params.replyOptions?.onItemEvent,
-        onCommandOutput: suppressAutomaticSourceDelivery
-          ? undefined
-          : params.replyOptions?.onCommandOutput,
-        onCompactionStart: suppressAutomaticSourceDelivery
-          ? undefined
-          : params.replyOptions?.onCompactionStart,
-        onCompactionEnd: suppressAutomaticSourceDelivery
-          ? undefined
-          : params.replyOptions?.onCompactionEnd,
         onToolResult: (payload: ReplyPayload) => {
+          markProgress();
           const run = async () => {
             markInboundDedupeReplayUnsafe();
             if (!suppressAutomaticSourceDelivery) {
@@ -1277,6 +1285,7 @@ export async function dispatchReplyFromConfig(
           return run();
         },
         onPlanUpdate: async (payload) => {
+          markProgress();
           markInboundDedupeReplayUnsafe();
           if (!suppressAutomaticSourceDelivery) {
             await onPlanUpdateFromReplyOptions?.(payload);
@@ -1287,6 +1296,7 @@ export async function dispatchReplyFromConfig(
           await sendPlanUpdate({ explanation: payload.explanation, steps: payload.steps });
         },
         onApprovalEvent: async (payload) => {
+          markProgress();
           markInboundDedupeReplayUnsafe();
           if (!suppressAutomaticSourceDelivery) {
             await onApprovalEventFromReplyOptions?.(payload);
@@ -1305,6 +1315,7 @@ export async function dispatchReplyFromConfig(
           await maybeSendWorkingStatus(label);
         },
         onPatchSummary: async (payload) => {
+          markProgress();
           markInboundDedupeReplayUnsafe();
           if (!suppressAutomaticSourceDelivery) {
             await onPatchSummaryFromReplyOptions?.(payload);
@@ -1319,6 +1330,7 @@ export async function dispatchReplyFromConfig(
           await maybeSendWorkingStatus(label);
         },
         onBlockReply: (payload: ReplyPayload, context?: BlockReplyContext) => {
+          markProgress();
           const run = async () => {
             if (
               payload.isReasoning !== true &&

src/config/schema.base.generated.ts

@@ -149,7 +149,7 @@ export const GENERATED_BASE_CONFIG_SCHEMA: BaseConfigSchemaResponse = {
             maximum: 9007199254740991,
             title: "Session Liveness Threshold (ms)",
             description:
-              "Age threshold in milliseconds for classifying long processing sessions as long-running, stalled, or stuck. Increase for long multi-tool turns; decrease for faster stale-session detection.",
+              "No-progress age threshold in milliseconds for classifying long processing sessions as long-running, stalled, or stuck. Reply, tool, status, block, and ACP progress reset the timer; repeated stuck diagnostics back off while unchanged.",
           },
           otel: {
             type: "object",
@@ -24485,7 +24485,7 @@ export const GENERATED_BASE_CONFIG_SCHEMA: BaseConfigSchemaResponse = {
     },
     "diagnostics.stuckSessionWarnMs": {
       label: "Session Liveness Threshold (ms)",
-      help: "Age threshold in milliseconds for classifying long processing sessions as long-running, stalled, or stuck. Increase for long multi-tool turns; decrease for faster stale-session detection.",
+      help: "No-progress age threshold in milliseconds for classifying long processing sessions as long-running, stalled, or stuck. Reply, tool, status, block, and ACP progress reset the timer; repeated stuck diagnostics back off while unchanged.",
       tags: ["observability", "storage"],
     },
     "diagnostics.otel.enabled": {

src/config/schema.help.ts

@@ -585,7 +585,7 @@ export const FIELD_HELP: Record<string, string> = {
   "diagnostics.enabled":
     "Master toggle for diagnostics instrumentation output in logs and telemetry wiring paths. Defaults to enabled; set false only in tightly constrained environments.",
   "diagnostics.stuckSessionWarnMs":
-    "Age threshold in milliseconds for classifying long processing sessions as long-running, stalled, or stuck. Increase for long multi-tool turns; decrease for faster stale-session detection.",
+    "No-progress age threshold in milliseconds for classifying long processing sessions as long-running, stalled, or stuck. Reply, tool, status, block, and ACP progress reset the timer; repeated stuck diagnostics back off while unchanged.",
   "diagnostics.otel.enabled":
     "Enables OpenTelemetry export pipeline for traces, metrics, and logs based on configured endpoint/protocol settings. Keep disabled unless your collector endpoint and auth are fully configured.",
   "diagnostics.otel.endpoint":

src/config/types.base.ts

@@ -276,7 +276,7 @@ export type DiagnosticsConfig = {
   enabled?: boolean;
   /** Optional ad-hoc diagnostics flags (e.g. "telegram.http"). */
   flags?: string[];
-  /** Threshold in ms before a processing session logs "stuck session" diagnostics. */
+  /** Threshold in ms before a processing session with no observed progress logs diagnostics. */
   stuckSessionWarnMs?: number;
   otel?: DiagnosticsOtelConfig;
   cacheTrace?: DiagnosticsCacheTraceConfig;

src/logging/diagnostic-session-state.ts

@@ -4,6 +4,7 @@ export type SessionState = {
   sessionId?: string;
   sessionKey?: string;
   lastActivity: number;
+  lastStuckWarnAgeMs?: number;
   state: SessionStateValue;
   queueDepth: number;
   toolCallHistory?: ToolCallRecord[];

src/logging/diagnostic.test.ts

@@ -25,6 +25,7 @@ import {
 import {
   logSessionStateChange,
   logMessageQueued,
+  markDiagnosticSessionProgress,
   resetDiagnosticStateForTest,
   resolveStuckSessionWarnMs,
   startDiagnosticHeartbeat,
@@ -202,6 +203,63 @@ describe("stuck session diagnostics threshold", () => {
     });
   });
 
+  it("does not warn while a processing session continues reporting progress", () => {
+    const events: DiagnosticEventPayload[] = [];
+    const unsubscribe = onDiagnosticEvent((event) => {
+      events.push(event);
+    });
+    try {
+      startDiagnosticHeartbeat({
+        diagnostics: {
+          enabled: true,
+          stuckSessionWarnMs: 30_000,
+        },
+      });
+      logSessionStateChange({ sessionId: "s1", sessionKey: "main", state: "processing" });
+      vi.advanceTimersByTime(45_000);
+      markDiagnosticSessionProgress({ sessionId: "s1", sessionKey: "main" });
+      vi.advanceTimersByTime(16_000);
+    } finally {
+      unsubscribe();
+    }
+
+    expect(events.filter((event) => event.type === "session.stuck")).toHaveLength(0);
+    expect(events.filter((event) => event.type === "session.stalled")).toHaveLength(0);
+    expect(events.filter((event) => event.type === "session.long_running")).toHaveLength(0);
+  });
+
+  it("backs off repeated stuck warnings while a session remains unchanged", () => {
+    const events: Array<{ ageMs?: number }> = [];
+    const recoverStuckSession = vi.fn();
+    const unsubscribe = onDiagnosticEvent((event) => {
+      if (event.type === "session.stuck") {
+        events.push({ ageMs: event.ageMs });
+      }
+    });
+    try {
+      startDiagnosticHeartbeat(
+        {
+          diagnostics: {
+            enabled: true,
+            stuckSessionWarnMs: 30_000,
+          },
+        },
+        { recoverStuckSession },
+      );
+      logSessionStateChange({ sessionId: "s1", sessionKey: "main", state: "processing" });
+      vi.advanceTimersByTime(91_000);
+      expect(events).toHaveLength(1);
+      expect(recoverStuckSession).toHaveBeenCalledTimes(1);
+
+      vi.advanceTimersByTime(30_000);
+    } finally {
+      unsubscribe();
+    }
+
+    expect(events.map((event) => event.ageMs)).toEqual([60_000, 120_000]);
+    expect(recoverStuckSession).toHaveBeenCalledTimes(2);
+  });
+
   it("reports active sessions as stalled instead of stuck when active work stops progressing", () => {
     const events: DiagnosticEventPayload[] = [];
     const recoverStuckSession = vi.fn();

src/logging/diagnostic.ts

@@ -416,6 +416,7 @@ export function logMessageQueued(params: {
   const state = getDiagnosticSessionState(params);
   state.queueDepth += 1;
   state.lastActivity = Date.now();
+  state.lastStuckWarnAgeMs = undefined;
   if (diag.isEnabled("debug")) {
     diag.debug(
       `message queued: sessionId=${state.sessionId ?? "unknown"} sessionKey=${
@@ -494,6 +495,7 @@ export function logSessionStateChange(
   const prevState = state.state;
   state.state = params.state;
   state.lastActivity = Date.now();
+  state.lastStuckWarnAgeMs = undefined;
   if (params.state === "idle") {
     state.queueDepth = Math.max(0, state.queueDepth - 1);
   }
@@ -518,6 +520,16 @@ export function logSessionStateChange(
   markActivity();
 }
 
+export function markDiagnosticSessionProgress(params: SessionRef) {
+  if (!areDiagnosticsEnabledForProcess()) {
+    return;
+  }
+  const state = getDiagnosticSessionState(params);
+  state.lastActivity = Date.now();
+  state.lastStuckWarnAgeMs = undefined;
+  markActivity();
+}
+
 function sessionAttentionFields(params: {
   classification: SessionAttentionClassification;
   activity: DiagnosticSessionActivitySnapshot;
@@ -562,6 +574,16 @@ export function logSessionAttention(
     activity,
     staleMs: params.thresholdMs,
   });
+  if (classification.eventType === "session.stuck") {
+    const nextWarnAgeMs =
+      state.lastStuckWarnAgeMs === undefined
+        ? params.thresholdMs
+        : Math.max(state.lastStuckWarnAgeMs + params.thresholdMs, state.lastStuckWarnAgeMs * 2);
+    if (params.ageMs < nextWarnAgeMs) {
+      return undefined;
+    }
+    state.lastStuckWarnAgeMs = params.ageMs;
+  }
   const label =
     classification.eventType === "session.stuck"
       ? "stuck session"