huangruiteng · Nov 6, 2026 · Nov 6, 2026 · Nov 6, 2026 · Nov 6, 2026 · Nov 6, 2026
@@ -178,3 +178,14 @@ assert.deepEqual(goalReplanCadenceEditorValue(null, enabledStrideDefault), enabl
 assert.equal(goalReplanCadenceEditorValue({stride_strategy: "off"}, enabledStrideDefault).stride_strategy, "off");
 assert.equal(goalReplanCadenceEditorValue({count_unit: "effective_turns", count: 3}, enabledStrideDefault).stride_strategy, "off");
 assert.equal(projectEditableCapabilityConfiguration(cadenceEditor, {}, enabledStrideDefault).stride_strategy, "evidence_conditioned_v1");
+
+// Independent Yardstick inheritance must not be pinned by editing the interval.
+const yardstickEffective = {...inheritedReview, yardstick_enabled: true};
+assert.equal(goalReplanCadenceEditorValue({count_unit: "effective_turns", count: 2}, yardstickEffective).yardstick_enabled, true);
+assert.deepEqual(goalReplanCadenceWriteValue({...yardstickEffective, count: 2}, true, false, false),
+  {...inheritedReview, count: 2}, "an unchanged inherited measurement choice stays live");
+for (const enabled of [true, false]) {
+  assert.deepEqual(goalReplanCadenceWriteValue({...yardstickEffective, yardstick_enabled: enabled}, false, false, true),
+    {stride_strategy: "off", yardstick_enabled: enabled}, "explicit measurement opt-in and opt-out survive form saving");
+}
+assert.equal(replanCadenceEditorDefaults(yardstickEffective, {...cadenceDefault, yardstick_enabled: false}).yardstick_enabled, true);
@@ -31,6 +31,9 @@ for (const copy of Object.values(deliveryReviewCopy)) {
     scope: { kind: "selected_work", todo_ids: ["todo_confirmed"] } }), snapshot.goal_id);
   assert.ok(deliveryReviewMarkdown(selected, copy).includes(copy.contract.selectedWork));
   assert.ok(deliveryReviewMarkdown(parseDeliveryReview(withContract(contract), snapshot.goal_id), copy).includes(copy.contract.allWork));
+  const goalOnly = parseDeliveryReview(withContract({ ...contract,
+    scope: { kind: "goal" }, tasks: [], held_todo_ids: [] }), snapshot.goal_id);
+  assert.ok(deliveryReviewMarkdown(goalOnly, copy).includes(copy.contract.goalOnly));
   const empty = parseDeliveryReview(withContract({ ...contract, criteria: [], tasks: [] }), snapshot.goal_id);
   assert.ok(deliveryReviewMarkdown(empty, copy).includes(copy.contract.noTasks));
   assert.ok(deliveryReviewMarkdown(empty, copy).includes(copy.contract.noCriteria));

@@ -5,8 +5,10 @@ import {mkdtemp, mkdir, readFile, writeFile, rm} from "node:fs/promises";
 import {tmpdir} from "node:os";
 import {resolve} from "node:path";
 import {createInterface} from "node:readline";
+import {createElement} from "react";
+import {renderToStaticMarkup} from "react-dom/server";
 import {resolveTestPython} from "../../../../scripts/test-python.mjs";
-import {readAnalysisSnapshot} from "../src/features/personal-workspace/goal-research-analysis.tsx";
+import {readAnalysisSnapshot, ResearchAnalysisPanel} from "../src/features/personal-workspace/goal-research-analysis.tsx";
 
 // A language-specific cache miss must not hide the same registered pack's
 // older, unspecified-language result or turn a read into model execution.
@@ -86,6 +88,25 @@ try {
   failOriginal = false;
   assert.deepEqual((await read()).originalPack, original, "The ordinary read retries recovery without execution");
 
+  // A failed/in-flight observation is unknown; only a successful empty read
+  // supports the absence message. Keep the existing read retry as recovery.
+  for (const zh of [true, false]) {
+    const empty = {...pack, availability: "available", selection: {exhaustive: false}, episode_count: 0};
+    const controller = {zh, pack: empty, requestPack: empty, location: {open: true}, panel: {current: null},
+      readOnly: true, loading: false, snapshot: {catalog: {}}, read: async () => {}, close: () => {}};
+    const noSaved = zh ? "此来源尚无已保存的分析。" : "This source has no saved analysis yet.";
+    const render = state => renderToStaticMarkup(createElement(ResearchAnalysisPanel, {analysis: {...controller, ...state}}));
+    assert.ok(render({}).includes(noSaved), "A successful empty source retains its ordinary absence state");
+    for (const state of [{loading: true}, {readError: "Source unavailable"}, {snapshot: {catalog: {}, originalReadError: "Original unavailable"}}]) {
+      const html = render(state);
+      assert.ok(!html.includes(noSaved), "An incomplete source read must not claim that no saved analysis exists");
+      if (!state.loading) {
+        assert.ok(html.includes('role="alert"'));
+        assert.ok(html.includes(zh ? 'aria-label="重试读取"' : 'aria-label="Retry reading"'));
+      }
+    }
+  }
+
   for (const options of [{current: () => false}, {signal: AbortSignal.abort()}]) {
     reset("zh");
     value = await read(true, undefined, options.signal, options.current);

@@ -78,6 +78,7 @@ export function replanCadenceEditorDefaults(effective: unknown, fallback: unknow
   if (Object.hasOwn(live, "count_unit")) defaults.count_unit = live.count_unit;
   if (Object.hasOwn(live, "count")) defaults.count = live.count;
   if (Object.hasOwn(live, "planning_horizon_goal")) defaults.planning_horizon_goal = live.planning_horizon_goal;
+  if (Object.hasOwn(live, "yardstick_enabled")) defaults.yardstick_enabled = live.yardstick_enabled;
   // Strategy omission keeps the existing default-off meaning.
   return defaults;
 }
@@ -102,13 +103,15 @@ export function goalReplanCadenceWriteValue(
   draft: Record<string, unknown>,
   cadenceOverride: boolean,
   horizonOverride = Object.hasOwn(draft, "planning_horizon_goal"),
+  yardstickOverride = Object.hasOwn(draft, "yardstick_enabled"),
 ): Record<string, unknown> {
   const value = { ...draft };
   if (!cadenceOverride) {
     delete value.count_unit;
     delete value.count;
   }
   if (!horizonOverride) delete value.planning_horizon_goal;
+  if (!yardstickOverride) delete value.yardstick_enabled;
   return value;
 }
 

@@ -1,3 +1,4 @@
+import {normalizeWorkYardstickCapture} from "../../../../../loopx/control_plane/work_items/yardstick_decision.js";
 import {sourceReadApiUrl} from "./local-status-query";
 import type {OracleEvidence} from "./goal-acceptance-observation";
 import { HANDOFF_MODES, EXECUTION_HANDOFF_MODES, type ExecutionHandoffMode } from "../../../../../loopx/control_plane/coordination/handoff_mode_vocabulary.js";
@@ -3008,6 +3009,7 @@ const exploreResultPageSchema = z.object({
     evidence_refs: z.array(z.string()), last_updated_at: z.string(),
     measurements: z.unknown().transform(value => normalizeExploreMeasurements(value)).optional(),
     measurement_error: z.string().optional(),
+    work_yardstick: z.unknown().transform(capture => normalizeWorkYardstickCapture({capture})).optional(),
     linked_todos_available: z.boolean().optional(),
     linked_todos: z.array(z.object({
       todo_id: z.string(), text: z.string(), status: z.string(), claimed_by: z.string(),
@@ -3019,5 +3021,8 @@ export async function fetchExploreResults(goalId: string, cursor?: string, readS
   const query = new URLSearchParams({goal_id: goalId, view: "findings", ...(cursor ? {cursor} : {})});
   const page = exploreResultPageSchema.parse(await requestJson<unknown>(sourceReadApiUrl(readSourceUrl, `/api/chat/explore-results?${query}`)));
   if (page.goal_id !== goalId) throw new Error("Explore result Goal mismatch");
+  for (const item of page.items) if (item.work_yardstick) normalizeWorkYardstickCapture({
+    capture: item.work_yardstick, goal_id: goalId, agent_id: item.agent_id, evidence_refs: item.evidence_refs,
+  });
   return page;
 }
@@ -167,6 +167,28 @@ assert.deepEqual(distinct.map(row => row.marker), ["stream", "foreign", "another
   "Equal text, other Sessions/Turns and separate follow-up identities remain distinct");
 console.log("conversation-returns: history-before-admission preserves the live identity and distinct replies");
 
+// A fast queued Turn can finish before its start/stream is observed. Its
+// canonical answer replaces the waiting receipt, never a live stream.
+const queuedReply = {sourceSessionId: "current", sourceTurnId: "same-turn", role: "assistant",
+  text: "", queued: true, pending: false};
+const completedQueue = reconcileConversationHistory([queuedReply], [storedRoles[1]], createHistory);
+assert.equal(completedQueue.length, 1);
+assert.equal(completedQueue[0].text, storedRoles[1].text);
+assert.equal(completedQueue[0].queued, false);
+assert.equal(completedQueue[0].sourceMessageId, "agent-message");
+assert.equal(reconcileConversationHistory(completedQueue, [storedRoles[1]], createHistory), completedQueue);
+const unboundQueuedReply = {...queuedReply, sourceTurnId: undefined, sourceMessageId: "agent-message"};
+const unboundAnswer = reconcileConversationReturns([unboundQueuedReply], "current", [{...storedRoles[1], turn_id: undefined}], createHistory);
+assert.equal(unboundAnswer[0].text, "");
+assert.equal(unboundAnswer[0].queued, true, "Two absent Turn identities do not prove queue execution");
+for (const change of [{pending: true, text: "Live fragment"}, {queued: false, text: "Existing answer"},
+  {sourceSessionId: "foreign"}, {sourceTurnId: "another-turn"}]) {
+  const original = {...queuedReply, ...change};
+  const observed = reconcileConversationHistory([original], [storedRoles[1]], createHistory);
+  assert.equal(observed.find(row => row === original || row.sourceTurnId === original.sourceTurnId
+    && row.sourceSessionId === original.sourceSessionId)?.text, original.text);
+}
+
 // A revised brief keeps its return obligation despite a delivered old conclusion.
 const revised = [{sourceSessionId: "original", collaborations: [
   {request_id: "a", request_revision: 2, returns: [{phase: "conclusion", status: "delivered", request_revision: 1}]},

@@ -9,6 +9,7 @@ type ConversationMessage = {
   sourceCreatedAt?: string;
   role?: string;
   pending?: boolean;
+  queued?: boolean;
   collaboration?: ChatVisibleMessage["collaboration"];
   collaborations?: ChatVisibleMessage["collaborations"];
   returnDelivery?: ChatVisibleMessage["return_delivery"];
@@ -81,16 +82,19 @@ export function reconcileConversationReturns<T extends ConversationMessage>(
       && source.queue_input_revision > (row.queueInputRevision ?? 0);
     const queueInputRevision = inputChanged ? source.queue_input_revision : row.queueInputRevision;
     const queueInputCancelled = inputChanged ? source.queue_input_cancelled : row.queueInputCancelled;
-    const text = inputChanged ? source.text : row.role === "user" && source.reply_context && row.text === source.text
+    const queueAnswered = row.queued === true && !row.pending && row.role === "assistant"
+      && Boolean(row.sourceTurnId) && source.role !== "user" && source.origin !== "manager_followup" && source.turn_id === row.sourceTurnId;
+    const queued = queueAnswered ? false : row.queued;
+    const text = inputChanged || queueAnswered ? source.text : row.role === "user" && source.reply_context && row.text === source.text
       ? source.reply_context.source_text : row.text;
     if (row.sourceMessageId === source.message_id && JSON.stringify(row.returnDelivery) === JSON.stringify(returnDelivery)
       && JSON.stringify(row.collaboration) === JSON.stringify(collaboration)
       && JSON.stringify(row.collaborations) === JSON.stringify(collaborations) && row.sourceCreatedAt === sourceCreatedAt
       && JSON.stringify(row.replyContext) === JSON.stringify(replyContext) && row.text === text
-      && row.queueInputRevision === queueInputRevision && row.queueInputCancelled === queueInputCancelled) return row;
+      && row.queueInputRevision === queueInputRevision && row.queueInputCancelled === queueInputCancelled && row.queued === queued) return row;
     changed = true;
     return { ...row, sourceMessageId: source.message_id, sourceCreatedAt, returnDelivery, collaboration, collaborations, replyContext, text,
-      queueInputRevision, queueInputCancelled };
+      queueInputRevision, queueInputCancelled, ...(queueAnswered ? {queued: false} : {}) };
   });
   for (const message of messages) {
     if (message.origin !== "manager_followup" || seen.has(message.message_id)) continue;

@@ -138,7 +138,7 @@ export function reviewNodeColumn(node: ReviewNode) {
 }
 
 type ConfiguredContract = GoalAcceptanceChecks;
-type ContractExportLabels = Record<"title" | "boundary" | "scope" | "selectedWork" | "allWork" | "source" | "revision" | "digest" | "objective" | "criteria"
+type ContractExportLabels = Record<"title" | "boundary" | "scope" | "goalOnly" | "selectedWork" | "allWork" | "source" | "revision" | "digest" | "objective" | "criteria"
   | "nonGoals" | "tasks" | "verification" | "unknown" | "noTasks" | "noCriteria" | "notApplicable"
   | "heldTasks" | "receipt" | "receiptNote" | "operation" | "verificationScope" | "allCriteria" | "passed" | "failed" | "exitCode"
   | "oracleChecker" | "oracleSourceDigest" | "oracleCheckedAt" | "oracleInputVersions" | "oracleOutputVersions", string> & {
@@ -236,7 +236,7 @@ export function deliveryReviewMarkdown(snapshot: DeliveryReviewSnapshot, labels:
   if (hasConfiguredGoalChecks(contract)) {
     const copy = labels.contract;
     rows.push("", `## ${copy.title}`, "", copy.boundary,
-      `${copy.scope}: ${contract.scope?.kind === "selected_work" ? copy.selectedWork : copy.allWork}`,
+      `${copy.scope}: ${contract.scope?.kind === "goal" ? copy.goalOnly : contract.scope?.kind === "selected_work" ? copy.selectedWork : copy.allWork}`,
       `${copy.source}: ${line(snapshot.goal_id)}`, `${copy.revision}: ${contract.revision}`, `${copy.digest}: ${line(contract.digest)}`,
       "", `### ${copy.objective}`, line(contract.objective || copy.unknown), "", `### ${copy.criteria}`);
     if (contract.non_goals.length) rows.push(`${copy.nonGoals}: ${line(contract.non_goals.join("; "))}`);
@@ -246,8 +246,8 @@ export function deliveryReviewMarkdown(snapshot: DeliveryReviewSnapshot, labels:
       if (criterion.oracle) rows.push(`  ${copy.oracleChecker}: ${line(criterion.oracle.source_id)} · ${copy.oracleState[criterion.oracle.state]}`,
         `  ${copy.oracleSourceDigest}: ${line(criterion.oracle.source_digest)}`);
     }
-    rows.push("", `### ${copy.tasks}`);
-    if (!contract.tasks.length) rows.push(copy.noTasks);
+    if (contract.tasks.length || contract.scope?.kind !== "goal") rows.push("", `### ${copy.tasks}`);
+    if (!contract.tasks.length && contract.scope?.kind !== "goal") rows.push(copy.noTasks);
     for (const task of contract.tasks) {
       rows.push(`- ${line(task.todo_id)}: ${copy.taskState[task.state]}`,
         `  ${copy.criteria}: ${line(task.criterion_ids.join(", ") || copy.unknown)}`);

@@ -22,6 +22,12 @@
       "Native and local measurement scopes remain distinct. Cancelled or invalid runs do not qualify a scientific comparison."
     ],
     "redactions": [],
+    "trajectory_archive": {
+      "url": "https://huggingface.co/datasets/huangrt01/loopx-edgebench-research-checkpoints-20261015/resolve/a97e88f951deb6827d7051a59927860e9e4c4476/20261107-real-benchmark-trajectories/ad_placement_optimization.zip",
+      "sha256": "45ce26e44a59587b5a086d3aabde9bba9899b184abd6a21029f75c8b6695a44b",
+      "manifest_url": "https://huggingface.co/datasets/huangrt01/loopx-edgebench-research-checkpoints-20261015/blob/a97e88f951deb6827d7051a59927860e9e4c4476/20261107-real-benchmark-trajectories/manifest.json",
+      "redactions_url": "https://huggingface.co/datasets/huangrt01/loopx-edgebench-research-checkpoints-20261015/blob/a97e88f951deb6827d7051a59927860e9e4c4476/20261107-real-benchmark-trajectories/redactions.json"
+    },
     "boundary": "Display-state events only. No authority or credential store, raw session transcript, hidden evaluator, workspace or model weights are included."
   },
   "source_counts": {

@@ -21,6 +21,12 @@
       "These runs include operational interventions and are not a qualified causal comparison. Local and native measurements retain separate scopes."
     ],
     "redactions": [],
+    "trajectory_archive": {
+      "url": "https://huggingface.co/datasets/huangrt01/loopx-edgebench-research-checkpoints-20261015/resolve/a97e88f951deb6827d7051a59927860e9e4c4476/20261107-real-benchmark-trajectories/grid_turing_robot.zip",
+      "sha256": "ea147c9ae7173e3f1ce9ca6e0bbda5e0103e2be035b8e6d0658dd35a2d282c8d",
+      "manifest_url": "https://huggingface.co/datasets/huangrt01/loopx-edgebench-research-checkpoints-20261015/blob/a97e88f951deb6827d7051a59927860e9e4c4476/20261107-real-benchmark-trajectories/manifest.json",
+      "redactions_url": "https://huggingface.co/datasets/huangrt01/loopx-edgebench-research-checkpoints-20261015/blob/a97e88f951deb6827d7051a59927860e9e4c4476/20261107-real-benchmark-trajectories/redactions.json"
+    },
     "boundary": "Display-state events only. No execution/authority store, credentials, raw session transcript, hidden evaluator, workspace or model weights are included."
   },
   "source_counts": {

@@ -22,6 +22,12 @@
       "Native and local measurement scopes remain distinct. Cancelled or invalid runs do not qualify a scientific comparison."
     ],
     "redactions": [],
+    "trajectory_archive": {
+      "url": "https://huggingface.co/datasets/huangrt01/loopx-edgebench-research-checkpoints-20261015/resolve/a97e88f951deb6827d7051a59927860e9e4c4476/20261107-real-benchmark-trajectories/schemathesis_config_modernization.zip",
+      "sha256": "e892d3b9569d84c7fff8ce80c2a0ae9f9a801d5f745eb980bfc918835327b2ea",
+      "manifest_url": "https://huggingface.co/datasets/huangrt01/loopx-edgebench-research-checkpoints-20261015/blob/a97e88f951deb6827d7051a59927860e9e4c4476/20261107-real-benchmark-trajectories/manifest.json",
+      "redactions_url": "https://huggingface.co/datasets/huangrt01/loopx-edgebench-research-checkpoints-20261015/blob/a97e88f951deb6827d7051a59927860e9e4c4476/20261107-real-benchmark-trajectories/redactions.json"
+    },
     "boundary": "Display-state events only. No authority or credential store, raw session transcript, hidden evaluator, workspace or model weights are included."
   },
   "source_counts": {