diff --git a/docs/BUG_HISTORY.md b/docs/BUG_HISTORY.md
index b41e426e..c532b44d 100644
--- a/docs/BUG_HISTORY.md
+++ b/docs/BUG_HISTORY.md
@@ -1398,3 +1398,18 @@
 - 防复发：只要首轮已向用户展示生时校正状态，可见文案必须同时呈现结构化候选边界；不得只在隐藏状态中保存范围，也不得为了润色首问重新引入模型调用。
 - 相关记录：BUG-055、BUG-067、BUG-072
 - 修复版本：待提交（本地可测）
+
+## BUG-075 | 生时校正强制每轮提问并阻碍开放叙事
+
+- 状态：resolved
+- 首次发现：2026-07-26
+- 最近更新：2026-07-26
+- 影响面：生时校正 Narrative Agent、非评分轮续接、确定性首轮、聊天消息完成态
+- 用户现象：用户叙述一段或多段人生经历后，Agent 会机械追加问题、反复继承上一轮追问，并在普通资料交流结束后显示“回答已完成”，整体表现像问卷收集器而不是能理解人生轨迹的校时助手。
+- 触发条件：非最终轮模型只做自然回应但同时返回结构化追问，或本轮没有生成新追问却沿用上一轮 `evidenceRequest`；所有 settled Assistant 消息都会渲染统一完成标签。
+- 根因：Narrative Prompt 强制每个非最终回复恰好以一个问题结束；服务端 repair 会把隐藏的 `evidenceRequest.prompt` 自动追加到正文；非评分轮会自动生成缺失信息追问并继承旧 `evidenceRequest`；通用消息组件把 settled 状态映射成“回答已完成”。
+- 修复：允许用户一次叙述一件或多件经历并自由连续表达，问题改为 Agent 可选行为；删除自动补问和已评分事件的聊天限制，普通自然回复使用 `evidenceRequest: null`；首轮只开放邀请叙述，非评分轮不再自动规划或继承问题，仅在明确的结构化纠正承接中保留必要状态；settled 消息不再渲染活动状态。事件抽取、保存、技术评分、候选收敛和分钟确认安全门保持不变。
+- 验证：Narrative Agent、Orchestrator 与聊天布局聚焦测试 101/101 通过；目标 ESLint 无错误或警告；Next.js 16 Webpack 生产构建与 `git diff --check` 通过。
+- 防复发：代码不得要求每轮必须提问，也不得把隐藏 prompt 自动拼接到 Agent 正文；没有可见短答问题时 `evidenceRequest` 应为 null，用户叙述节奏由 Agent 与用户共同决定，代码只维护记录、抽取、评分、收敛与确认边界。
+- 相关记录：BUG-070、BUG-072、BUG-074
+- 修复版本：待提交（本地可测）
diff --git a/frontend/src/components/agent-activity-status.tsx b/frontend/src/components/agent-activity-status.tsx
index 0c1c422a..11d9b03e 100644
--- a/frontend/src/components/agent-activity-status.tsx
+++ b/frontend/src/components/agent-activity-status.tsx
@@ -1,4 +1,3 @@
-import { CircleCheck } from "lucide-react";
 import { ThinkingOrb, type OrbState } from "thinking-orbs";
 
 const labels = {
@@ -8,10 +7,9 @@ const labels = {
   listening: "正在聆听…",
   composing: "正在组织回答…",
   shaping: "正在生成结果…",
-  completed: "回答已完成",
-} as const satisfies Record<OrbState | "completed", string>;
+} as const satisfies Record<OrbState, string>;
 
-export type AgentActivityState = OrbState | "completed";
+export type AgentActivityState = OrbState;
 
 export function AgentActivityStatus({
   state,
@@ -21,10 +19,8 @@ export function AgentActivityStatus({
   label?: string;
 }>) {
   return (
-    <div className="agent-activity-status" role={state === "completed" ? undefined : "status"}>
-      {state === "completed"
-        ? <CircleCheck aria-hidden="true" size={20} strokeWidth={1.8} />
-        : <ThinkingOrb aria-hidden="true" state={state} size={20} />}
+    <div className="agent-activity-status" role="status">
+      <ThinkingOrb aria-hidden="true" state={state} size={20} />
       <span>{label}</span>
     </div>
   );
diff --git a/frontend/src/components/chat-message-row.tsx b/frontend/src/components/chat-message-row.tsx
index 6103538d..82537dbf 100644
--- a/frontend/src/components/chat-message-row.tsx
+++ b/frontend/src/components/chat-message-row.tsx
@@ -51,14 +51,12 @@ export function ChatMessageRow({ message }: { readonly message: ChatMessageView
         <div className="message-bubble">
           {message.role === "assistant" ? (
             <>
-              <AgentActivityStatus
-                state={message.state === "thinking"
-                  ? "working"
-                  : message.state === "streaming"
-                    ? "composing"
-                    : "completed"}
-                label={message.state === "thinking" ? "正在核对星盘信息…" : undefined}
-              />
+              {message.state !== "settled" && (
+                <AgentActivityStatus
+                  state={message.state === "thinking" ? "working" : "composing"}
+                  label={message.state === "thinking" ? "正在核对星盘信息…" : undefined}
+                />
+              )}
               {message.text && <ChatMessageContent text={message.text} />}
             </>
           ) : <p>{message.text}</p>}
diff --git a/frontend/src/lib/conversational-rectification/narrative-agent.ts b/frontend/src/lib/conversational-rectification/narrative-agent.ts
index 78470183..26f8e3d9 100644
--- a/frontend/src/lib/conversational-rectification/narrative-agent.ts
+++ b/frontend/src/lib/conversational-rectification/narrative-agent.ts
@@ -65,7 +65,6 @@ const labeledYearChoicesPattern = /A\s*[.、:：)]?[\s\S]{0,80}(?:19|20)\d{2}\s*
 const affirmativeAnswerPattern = /^\s*(?:是(?:的)?|对(?:的)?|没错|正确|确认|就是|嗯+|没问题)\s*[。.!！,，]?\s*$/u;
 const negativeAnswerPattern = /^\s*(?:不是|不对|错了|并不是|否)\s*[。.!！,，]?\s*$/u;
 const proposedDateQuestionPattern = /(?:19|20)\d{2}\s*年(?:\s*(?:1[0-2]|0?[1-9])\s*月)?(?:\s*(?:3[01]|[12]\d|0?[1-9])\s*(?:日|号))?[\s\S]{0,30}(?:吗|是否|是不是|确认|对不对|正确)/u;
-const nonScoringDetailPattern = /为什么|原因|主动|被动|自愿|被迫|影响|感受|具体(?:体现|情况|经过)|哪些方面|正式工作|实习|兼职/u;
 const domainLabels = {
   career: "事业",
   education: "学业",
@@ -301,17 +300,6 @@ export function validateNarrativeAgainstPacket(
       if (!allowedDomains.has(domain)) issues.push(`evidence domain ${domain} is not packet-grounded`);
     }
     const followUp = output.evidenceRequest.followUp;
-    if (followUp?.kind === "event_detail"
-      && packet.scoredHistoricalEvidence.some((item) => item.evidenceId === followUp.evidenceId)) {
-      issues.push("event detail follow-up targets already scored evidence");
-    }
-    const latestActiveEvidenceId = context.eventLedger?.filter((item) => item.active).at(-1)?.id;
-    if (followUp?.kind === "new_event"
-      && latestActiveEvidenceId
-      && nonScoringDetailPattern.test(output.evidenceRequest.prompt)
-      && packet.scoredHistoricalEvidence.some((item) => item.evidenceId === latestActiveEvidenceId)) {
-      issues.push("new-event follow-up disguises detail about already scored evidence");
-    }
     if (proposedDateQuestionPattern.test(output.evidenceRequest.prompt)
       && (followUp?.kind !== "event_date"
         || followUp.answerMode !== "yes_no"
@@ -454,45 +442,6 @@ function boundedReceiptIssues(issues: readonly string[]): string[] {
     .map((issue) => issue.trim().slice(0, 240) || "narrative_mismatch");
 }
 
-function ensureSentence(value: string, sentence: string): string {
-  const trimmed = value.trim();
-  return trimmed ? `${trimmed}\n${sentence}` : sentence;
-}
-
-function endsWithIncompletePrompt(value: string): boolean {
-  return /(?:[-—–:：,，、]|\.\.\.|…|我需要(?:确认|了解|知道)|关键信息)\s*$/.test(value.trim());
-}
-
-function hasExplicitQuestion(value: string): boolean {
-  return /[？?]/.test(value);
-}
-
-function repairRequiredSafetyLanguage(
-  output: RectificationNarrativeModelOutput,
-  phase: RectificationNarrativePhase,
-): RectificationNarrativeModelOutput {
-  let narrative = output.narrative;
-  const evidenceRequest = output.evidenceRequest;
-  const modelEvidencePrompt = output.evidenceRequest?.prompt.trim() ?? "";
-
-  // Every collecting turn must end with one visible, model-authored question.
-  // evidenceRequest.prompt is generated in the same model call, so appending it
-  // repairs truncated or analysis-only prose without introducing a business
-  // template or changing the Agent's chosen conversational direction.
-  if (phase !== "final" && evidenceRequest && (
-    endsWithIncompletePrompt(narrative)
-    || !hasExplicitQuestion(narrative)
-  )) {
-    narrative = ensureSentence(narrative, modelEvidencePrompt || evidenceRequest.prompt);
-  }
-
-  return {
-    ...output,
-    narrative,
-    evidenceRequest,
-  };
-}
-
 type NarrativeDiagnosticIssueCode =
   | "candidate_status_mismatch"
   | "representative_time_mismatch"
@@ -560,11 +509,11 @@ function promptFor(
 ): string {
   if (phase === "first") {
     return JSON.stringify({
-      task: "用自然、有人味的中文开启生时校正；简短回应当前范围，再只问一个最有信息量的问题。不要使用固定模板。",
+      task: "用自然、有人味的中文开启生时校正。用户可以按自己的节奏自由叙述，一次说一件或多件经历；自然回应即可，不必每轮提问，也不要使用固定模板。",
       phase,
       conversationContext: narrativeConversationContext(context),
       packet: grounding(packet, phase),
-      output: "只返回 narrative，以及 evidenceRequest。evidenceRequest 可用 domains 作为不可见路由元数据，并包含 datePrecision、prompt、followUp；narrative 不得输出或讨论事件分类、领域标签，也不要重复输出候选状态、时间、分盘或引用字段。",
+      output: "只返回 narrative 和 evidenceRequest。没有提出需要下一轮短答承接的明确问题时，evidenceRequest 必须为 null；若提出明确问题，可用 domains 作为不可见路由元数据，并包含 datePrecision、prompt、followUp。narrative 不得输出或讨论事件分类、领域标签，也不要重复输出候选状态、时间、分盘或引用字段。",
       safety: "技术事实只能来自 packet；不能确认未经验证的分钟；不得展示内部权重、分数、事件分类或内部路由元数据。",
       retryIssues: boundedReceiptIssues(retryIssues),
     });
@@ -592,21 +541,21 @@ function promptFor(
       internalEventDomainsAndRoutingMustNeverBeShown: true,
     },
     conversationGuidance: {
-      preferOneHighInformationQuestion: phase !== "final",
-      everyNonFinalReplyMustEndWithExactlyOneVisibleQuestion: phase !== "final",
-      respondToLatestEvidenceBeforeAsking: phase === "intermediate",
+      freeConversation: phase !== "final",
+      questionsAreOptional: phase !== "final",
+      userControlsNarrativePace: phase !== "final",
+      acceptMultipleEventsInOneMessage: phase !== "final",
+      acknowledgeAndReflectBeforeAnyClarification: phase === "intermediate",
+      doNotTurnEveryMessageIntoAQuestionnaire: phase !== "final",
+      askOnlyWhenAClarificationWouldMateriallyHelpTheConversation: phase !== "final",
       neverMentionHowTheEventWasClassifiedOrLabeledInternally: phase === "intermediate",
       doNotVolunteerNotEvaluatedOrBlockedTechniqueInventory: phase === "intermediate",
       doNotRepeatCandidateBoundaryUnlessItChangedOrTheUserAsked: phase === "intermediate",
-      continueCurrentEventWhenItRemainsInformative: phase === "intermediate",
       resolveDateContradictionsBeforeScoring: phase === "intermediate",
       mergeSameEventDetailsWithoutDoubleCounting: phase === "intermediate",
-      onlyAskForDateEventIdentityOrInformationThatCanChangeTheScoringDomain: phase === "intermediate",
-      doNotAskWhyWhetherVoluntaryOrWhatImpactForAlreadyScoreableEvidence: phase === "intermediate",
       useEventLedgerToAvoidRepeatingAnsweredQuestions: phase === "intermediate",
-      askForDatesOnlyWhenNeededToIdentifyOrScoreTheEvent: phase !== "final",
-      persistFollowUpState: phase !== "final"
-        ? "Persist the exact question intent. Use event_detail or event_date with an existing evidenceId. For a yes/no date proposal, set answerMode=yes_no and proposedDate={value,precision}; otherwise use answerMode=free_text and no proposedDate. Use new_event with null evidenceId only for a genuinely new event."
+      optionalFollowUpState: phase !== "final"
+        ? "Only when narrative contains a clear question that expects a short next-turn answer, persist its exact intent. Use event_detail or event_date with an existing evidenceId. For a yes/no date proposal, set answerMode=yes_no and proposedDate={value,precision}; otherwise use answerMode=free_text and no proposedDate. Use new_event with null evidenceId only for a genuinely new event. Otherwise evidenceRequest must be null."
         : false,
       boundedResultBoundary: phase === "final" && packet.candidate.status === "pending_validation"
         ? "当前只支持候选范围，系统验证尚未闭环。本次不会替换当前排盘时间，也不再要求用户继续提供人生事件；evidenceRequest 必须为 null。"
@@ -619,13 +568,11 @@ function promptFor(
 
 function fallbackNarrative(packet: RectificationTechnicalPacket, phase: RectificationNarrativePhase): string {
   const candidate = packet.candidate;
-  const nextDomain = packet.suggestedDomains[0]?.domain;
-  const nextLabel = nextDomain ? domainLabels[nextDomain] : "重要经历";
   const phaseLine = phase === "final" && candidate.status === "ready_for_confirmation"
     ? "当前证据已形成候选总结，但仍有残余不确定性；只有明确确认后才会替换当前排盘时间。"
     : phase === "final"
       ? "当前证据只能支持候选范围，系统验证尚未闭环；本次不会替换当前排盘时间，也不再强制追问更多人生事件。"
-    : `先说一件已经发生的${nextLabel}事件好吗？尽量写明哪一年、哪一月以及发生了什么。`;
+    : `我收到了这段叙述。你可以继续讲这段经历，也可以按自己的节奏说下一件想到的事。`;
   return [
     `当前仍在核对 ${candidate.range.startTime}–${candidate.range.endTime} 的候选范围，还不能把其中某一分钟当作确定出生时间。`,
     phaseLine,
@@ -647,12 +594,7 @@ function fallbackOutput(
     sensitiveLayers: packet.sensitiveLayers.map((item) => item.layer),
     referenceIds: [],
     domainReasons: packet.suggestedDomains.map((item) => ({ ...item })),
-    evidenceRequest: phase === "final" ? null : {
-      domains: packet.suggestedDomains.slice(0, 4).map((item) => item.domain),
-      datePrecision: "month_preferred",
-      prompt: "请提供已经发生的真实事件，并尽量写明哪一年、哪一月以及发生了什么。",
-      followUp: { kind: "new_event", evidenceId: null },
-    },
+    evidenceRequest: null,
   };
 }
 
@@ -798,10 +740,7 @@ export async function generateRectificationNarrative(input: {
         issues,
       ), { signal, attempt });
       const modelId = modelIdSchema.parse(generated.modelId ?? defaultModelId);
-      const output = repairRequiredSafetyLanguage(
-        parseModelOutput(generated.text, input.packet),
-        input.phase,
-      );
+      const output = parseModelOutput(generated.text, input.packet);
       const validation = validateNarrativeAgainstPacket(
         output,
         input.packet,
diff --git a/frontend/src/lib/conversational-rectification/orchestrator.ts b/frontend/src/lib/conversational-rectification/orchestrator.ts
index 5527a344..19ce389d 100644
--- a/frontend/src/lib/conversational-rectification/orchestrator.ts
+++ b/frontend/src/lib/conversational-rectification/orchestrator.ts
@@ -603,7 +603,7 @@ function openingRectificationState(input: {
     journeyProtocol: "conversational-evidence-v3",
     status: "active",
     turnVersion: 0,
-    narrative: `根据你填写的出生时间信息，当前先核对 ${range.startTime}–${range.endTime}。这只是待核对范围，还不能把其中某一分钟当作已确认出生时间。请先说一件时间最明确、影响比较大的真实经历，并尽量告诉我发生的年月。`,
+    narrative: `根据你填写的出生时间信息，当前先核对 ${range.startTime}–${range.endTime}。这只是待核对范围，还不能把其中某一分钟当作已确认出生时间。你可以按自己的节奏讲已经发生的人生经历，一次说一件或连续说多件都可以；记得的年月可以自然地带上，不确定也没关系。`,
     candidate: {
       status: "pending_validation",
       representativeTime,
@@ -616,13 +616,7 @@ function openingRectificationState(input: {
       sensitiveLayers: [],
       candidateDifferenceRefs: [],
     },
-    evidenceRequest: {
-      domains: ["career", "education", "relocation", "relationship"],
-      datePrecision: "month_preferred",
-      freeTextAllowed: true,
-      prompt: "请说一件已经发生、时间比较明确的重要经历，并告诉我大约是哪一年、哪一月？",
-      followUp: { kind: "new_event", evidenceId: null },
-    },
+    evidenceRequest: null,
     evidenceRecap: [],
     actions: ["answer", "pause", "abandon"],
     pendingConsultationQuestion: input.pendingConsultationQuestion,
@@ -659,45 +653,30 @@ function nonScoringTurn(input: {
   }>;
 }): { readonly turn: ConversationalRectificationTurn; readonly receipt: ValidationReceipt } {
   const allEvidence = [...input.current.eventEvidence, ...input.newEvidence];
-  const latestIncomplete = input.newEvidence
-    .filter((item) => item.extractionStatus === "needs_clarification")
-    .at(-1);
   const authoredNarrative = input.authoredNarrative;
   const latestSummary = input.newEvidence.at(-1)?.eventSummary;
   const fallbackSubject = latestSummary && latestSummary !== "事件内容待补充"
     ? latestSummary
     : input.latestUserText.trim().slice(0, 80);
   const narrative = authoredNarrative?.narrative
-    ?? `我收到了你这轮关于“${fallbackSubject || "这段经历"}”的补充，但这次分析暂时没有完成。内容会保留，你可以继续补充它的时间和经过，或直接说下一件已经发生的经历。`;
+    ?? `我收到了你这轮关于“${fallbackSubject || "这段经历"}”的补充，内容已经保留。你可以继续讲这段经历，也可以按自己的节奏说下一件想到的事。`;
   const status = input.correctionReset
     ? "active" as const
     : input.current.status === "confirming" ? "confirming" as const : "active" as const;
   const actions = actionsFor(status);
-  const clarificationFollowUp = latestIncomplete?.dateValue === null
-    && latestIncomplete.eventSummary !== "事件内容待补充"
-    ? { kind: "event_date" as const, evidenceId: latestIncomplete.id }
-    : latestIncomplete?.dateValue
-      && latestIncomplete.eventSummary === "事件内容待补充"
-      ? { kind: "event_detail" as const, evidenceId: latestIncomplete.id }
+  const authoredRequest = authoredNarrative?.output.evidenceRequest;
+  const priorRequest = input.current.latestTurn.evidenceRequest;
+  const evidenceRequest = authoredRequest
+    ? {
+        domains: authoredRequest.domains,
+        datePrecision: authoredRequest.datePrecision,
+        freeTextAllowed: true as const,
+        prompt: authoredRequest.prompt,
+        followUp: input.followUpOverride ?? authoredRequest.followUp,
+      }
+    : input.followUpOverride && priorRequest
+      ? { ...priorRequest, followUp: input.followUpOverride }
       : null;
-    const authoredRequest = authoredNarrative?.output.evidenceRequest;
-    const priorRequest = input.current.latestTurn.evidenceRequest;
-  const evidenceRequest = status === "confirming" && priorRequest === null
-    ? null
-    : authoredRequest
-      ? {
-          domains: authoredRequest.domains,
-          datePrecision: authoredRequest.datePrecision,
-          freeTextAllowed: true as const,
-          prompt: authoredRequest.prompt,
-          followUp: input.followUpOverride ?? authoredRequest.followUp,
-        }
-      : priorRequest
-        ? {
-            ...priorRequest,
-            followUp: input.followUpOverride ?? clarificationFollowUp ?? priorRequest.followUp,
-          }
-        : null;
   const parsed = conversationalRectificationTurnSchema.safeParse({
     ...input.current.latestTurn,
     status,
diff --git a/frontend/tests/chat-stream-layout.test.ts b/frontend/tests/chat-stream-layout.test.ts
index de13c7ed..462c0e0a 100644
--- a/frontend/tests/chat-stream-layout.test.ts
+++ b/frontend/tests/chat-stream-layout.test.ts
@@ -47,12 +47,12 @@ test("does not duplicate a completed assistant answer while loading state settle
 test("shows honest agent activity states before and during streamed text", () => {
   assert.match(messageRowSource, /message\.state === "thinking"[\s\S]*?\? "working"/);
   assert.match(messageRowSource, /message\.state === "thinking" \? "正在核对星盘信息…"/);
-  assert.match(messageRowSource, /message\.state === "streaming"[\s\S]*?"composing"/);
-  assert.match(messageRowSource, /: "completed"/);
+  assert.match(messageRowSource, /message\.state !== "settled"/);
+  assert.match(messageRowSource, /message\.state === "thinking" \? "working" : "composing"/);
   assert.match(messageRowSource, /message\.text && <ChatMessageContent text=\{message\.text\}/);
   assert.match(activitySource, /<ThinkingOrb aria-hidden="true" state=\{state\} size=\{20\}/);
-  assert.match(activitySource, /<CircleCheck aria-hidden="true" size=\{20\}/);
-  assert.match(activitySource, /satisfies Record<OrbState \| "completed", string>/);
+  assert.doesNotMatch(activitySource, /CircleCheck|回答已完成|completed/);
+  assert.doesNotMatch(messageRowSource, /: "completed"/);
   assert.doesNotMatch(globalStyles, /\.thinking\b/);
 });
 
diff --git a/frontend/tests/conversational-narrative-agent.test.ts b/frontend/tests/conversational-narrative-agent.test.ts
index 4c431f8d..7ddde858 100644
--- a/frontend/tests/conversational-narrative-agent.test.ts
+++ b/frontend/tests/conversational-narrative-agent.test.ts
@@ -144,23 +144,18 @@ test("preserves a complete model-authored first-turn narrative verbatim", async
   assert.equal(result.output.evidenceRequest?.prompt, "那段关系大约是什么时候发生的？");
 });
 
-test("completes a first-turn introduction with the model-authored question", async () => {
+test("allows a first-turn acknowledgement without forcing a question", async () => {
+  const narrative = "我们先把你记得的出生范围当作起点。你可以按自己的节奏往下讲。";
   const result = await generateRectificationNarrative({
     phase: "first",
     packet: syntheticTechnicalPacket(),
-    generator: generator([{
-      narrative: "我们先把你记得的出生范围当作起点，再用真实经历逐步核对。",
-      evidenceRequest: {
-        domains: ["relationship"],
-        datePrecision: "month_preferred",
-        prompt: "先说一件时间比较明确的关系转折，好吗？",
-      },
-    }]),
+    generator: generator([{ narrative, evidenceRequest: null }]),
   });
 
   assert.equal(result.fallbackUsed, false);
-  assert.match(result.narrative, /真实经历逐步核对。[\s\S]*关系转折，好吗？/);
-  assert.equal((result.narrative.match(/[？?]/g) ?? []).length, 1);
+  assert.equal(result.narrative, narrative);
+  assert.equal(result.output.evidenceRequest, null);
+  assert.doesNotMatch(result.narrative, /[？?]/);
 });
 
 test("repairs a natural intermediate follow-up prompt without discarding the specific reply", async () => {
@@ -226,73 +221,18 @@ test("preserves the target event in an intermediate detail follow-up", async ()
   });
 });
 
-test("appends the model-authored question when an intermediate reply only acknowledges the event", async () => {
-  const output = {
-    ...richOutput(),
-    narrative: "已记录本科毕业后衔接读研，这是一段连续教育转折，暂时不重复计数。",
-    evidenceRequest: {
-      domains: ["career" as const],
-      datePrecision: "month_preferred" as const,
-      prompt: "毕业后的第一份工作是什么时候开始的？",
-    },
-  };
-
+test("keeps a natural intermediate acknowledgement without appending a question", async () => {
+  const narrative = "已记录本科毕业后衔接读研，这是一段连续教育转折，暂时不重复计数。你可以继续讲。";
   const result = await generateRectificationNarrative({
     phase: "intermediate",
     packet: syntheticTechnicalPacket(),
-    generator: generator([output]),
+    generator: generator([{ narrative, evidenceRequest: null }]),
   });
 
   assert.equal(result.fallbackUsed, false);
-  assert.match(result.narrative, /连续教育转折[\s\S]*第一份工作是什么时候开始的？/);
-  assert.equal((result.narrative.match(/[？?]/g) ?? []).length, 1);
-});
-
-test("repairs a narrative that stops at a dangling question introduction", async () => {
-  const output = {
-    ...richOutput(),
-    narrative: [
-      "谢谢你补充了当时的压力感受，这让这段经历更完整。",
-      "为了判断这段经历的具体节奏，我需要确认一个关键信息——",
-    ].join("\n\n"),
-    evidenceRequest: {
-      domains: ["career" as const],
-      datePrecision: "month_preferred" as const,
-      prompt: "你感到压力最强的时候，大约是哪一年、哪一个月？",
-    },
-  };
-
-  const result = await generateRectificationNarrative({
-    phase: "intermediate",
-    packet: syntheticTechnicalPacket(),
-    generator: generator([output]),
-  });
-
-  assert.equal(result.fallbackUsed, false);
-  assert.match(result.narrative, /关键信息——[\s\S]*压力最强的时候[\s\S]*哪一年、哪一个月/);
-  assert.equal((result.narrative.match(/[？?]/g) ?? []).length, 1);
-});
-
-test("turns a complete intermediate acknowledgement into a visible one-question turn", async () => {
-  const output = {
-    ...richOutput(),
-    narrative: "这次入职会作为一条事业事件记录，但还需要知道后续是否发生过离职、转岗或升职。",
-    evidenceRequest: {
-      domains: ["career" as const],
-      datePrecision: "month_preferred" as const,
-      prompt: "这份工作后来第一次发生明确变化是在什么时候？",
-    },
-  };
-
-  const result = await generateRectificationNarrative({
-    phase: "intermediate",
-    packet: syntheticTechnicalPacket(),
-    generator: generator([output]),
-  });
-
-  assert.equal(result.fallbackUsed, false);
-  assert.match(result.narrative, /离职、转岗或升职。[\s\S]*第一次发生明确变化是在什么时候？/);
-  assert.equal((result.narrative.match(/[？?]/g) ?? []).length, 1);
+  assert.equal(result.narrative, narrative);
+  assert.equal(result.output.evidenceRequest, null);
+  assert.doesNotMatch(result.narrative, /[？?]/);
 });
 
 test("hides an uninvoked technique inventory during evidence collection", async () => {
@@ -476,7 +416,8 @@ test("passes the user's latest concrete event to an intermediate skill-guided re
   });
 
   assert.match(prompts[0] ?? "", /离开家乡去上海开始第一份长期工作/);
-  assert.match(prompts[0] ?? "", /respondToLatestEvidenceBeforeAsking/);
+  assert.match(prompts[0] ?? "", /acknowledgeAndReflectBeforeAnyClarification/);
+  assert.match(prompts[0] ?? "", /questionsAreOptional/);
   assert.match(prompts[0] ?? "", /doNotRepeatCandidateBoundaryUnlessItChangedOrTheUserAsked/);
 });
 
@@ -555,7 +496,9 @@ test("passes the active event ledger and unresolved facts to the intermediate ag
   assert.match(prompt, /23年关系结束后发生过一次交通事故/);
   assert.match(prompt, /2024-08-08/);
   assert.match(prompt, /一段重要关系结束/);
-  assert.match(prompt, /continueCurrentEventWhenItRemainsInformative/);
+  assert.match(prompt, /freeConversation/);
+  assert.match(prompt, /questionsAreOptional/);
+  assert.match(prompt, /acceptMultipleEventsInOneMessage/);
   assert.match(prompt, /resolveDateContradictionsBeforeScoring/);
   assert.match(prompt, /mergeSameEventDetailsWithoutDoubleCounting/);
 });
@@ -868,7 +811,7 @@ test("rejects a follow-up that still targets evidence completed by an affirmativ
   assert.ok(result.issues.includes("resolved follow-up still targets completed evidence"));
 });
 
-test("rejects non-scoring detail questions for evidence already used by the scorer", () => {
+test("allows natural discussion of an event after it has contributed to scoring", () => {
   const evidenceId = "11111111-1111-4111-8111-111111111111";
   const packet: RectificationTechnicalPacket = {
     ...syntheticTechnicalPacket(),
@@ -882,7 +825,7 @@ test("rejects non-scoring detail questions for evidence already used by the scor
   };
   const output = richOutput();
   assert.ok(output.evidenceRequest);
-  const invalid = {
+  const conversational = {
     ...output,
     evidenceRequest: {
       ...output.evidenceRequest,
@@ -896,10 +839,10 @@ test("rejects non-scoring detail questions for evidence already used by the scor
     },
   } satisfies RectificationNarrativeModelOutput;
 
-  const result = validateNarrativeAgainstPacket(invalid, packet, "intermediate");
-
-  assert.equal(result.valid, false);
-  assert.ok(result.issues.includes("event detail follow-up targets already scored evidence"));
+  assert.deepEqual(validateNarrativeAgainstPacket(conversational, packet, "intermediate"), {
+    valid: true,
+    issues: [],
+  });
 });
 
 test("retries a grounded validation failure once with a compact packet", async () => {
@@ -1042,7 +985,7 @@ test("retries when a final narrative asks for more evidence", async () => {
   assert.equal(result.output.evidenceRequest, null);
 });
 
-test("retries a scored-event detail question mislabeled as new_event", async () => {
+test("does not force a retry when the Agent naturally discusses a scored event", async () => {
   const evidenceId = "00000000-0000-4000-8000-000000000709";
   const packet = {
     ...syntheticTechnicalPacket(),
@@ -1054,23 +997,10 @@ test("retries a scored-event detail question mislabeled as new_event", async ()
       ruleRefs: ["synthetic-education-rule"],
     }],
   };
-  const invalid = {
+  const output = {
     ...richOutput(),
-    evidenceRequest: {
-      domains: ["career" as const],
-      datePrecision: "month_preferred" as const,
-      prompt: "这几个月里，学业压力具体体现在哪些方面，主要原因是什么？",
-      followUp: { kind: "new_event" as const, evidenceId: null },
-    },
-  };
-  const valid = {
-    ...richOutput(),
-    evidenceRequest: {
-      domains: ["career" as const],
-      datePrecision: "month_preferred" as const,
-      prompt: "请再说一件已经发生的事业变化，并写明哪一年、哪一月。",
-      followUp: { kind: "new_event" as const, evidenceId: null },
-    },
+    narrative: "我理解，这几个月的学业压力不只是结果，也影响了你当时的选择。",
+    evidenceRequest: null,
   };
   const result = await generateRectificationNarrative({
     phase: "intermediate",
@@ -1087,12 +1017,13 @@ test("retries a scored-event detail question mislabeled as new_event", async ()
         correctsEvidenceIds: [],
       }],
     },
-    generator: generator([invalid, valid]),
+    generator: generator([output]),
   });
 
-  assert.equal(result.attempts, 2);
+  assert.equal(result.attempts, 1);
   assert.equal(result.fallbackUsed, false);
-  assert.equal(result.output.evidenceRequest?.prompt, valid.evidenceRequest.prompt);
+  assert.equal(result.output.evidenceRequest, null);
+  assert.equal(result.narrative, output.narrative);
 });
 
 test("records first-turn generation timeouts separately from schema failures", async () => {
diff --git a/frontend/tests/conversational-rectification-orchestrator.test.ts b/frontend/tests/conversational-rectification-orchestrator.test.ts
index e39ae2ad..2788abb5 100644
--- a/frontend/tests/conversational-rectification-orchestrator.test.ts
+++ b/frontend/tests/conversational-rectification-orchestrator.test.ts
@@ -106,6 +106,7 @@ function validGenerator(
   detailQuestion = "这份工作是正式工作，还是实习或兼职？",
   mislabelLatestDetailAsNewEvent = false,
   classifyEvidenceDomain?: (text: string) => "career" | "education" | "finance" | "health_pressure" | "relocation" | "relationship" | "family" | "other" | null,
+  freeNarrativeFromGeneration?: number,
 ) {
   let generation = 0;
   return {
@@ -161,7 +162,12 @@ function validGenerator(
       const asksForLatestDetail = continueLatestEvent
         && request.phase === "intermediate"
         && latestActiveEvent !== undefined;
-      const narrative = [
+      const freeNarrative = request.phase !== "final"
+        && freeNarrativeFromGeneration !== undefined
+        && generation >= freeNarrativeFromGeneration;
+      const narrative = freeNarrative
+        ? "我明白，这段经历已经记下。你可以继续按自己的节奏讲。"
+        : [
         request.phase === "intermediate" && latest
           ? `记下了：${latest.dateLabel} · ${latest.summary}。`
           : `当前仍在核对 ${value.candidate.rangeStart}–${value.candidate.rangeEnd} 的候选范围，不能视为已经确认的出生分钟。`,
@@ -173,7 +179,7 @@ function validGenerator(
           : asksForLatestDetail
             ? detailQuestion
             : `先说一件已经发生的${nextDomain}经历好吗？请写明哪一年、哪一月以及发生了什么。`,
-      ].join("");
+        ].join("");
       return { text: JSON.stringify({
         narrative,
         candidateStatus: value.candidate.status,
@@ -185,7 +191,7 @@ function validGenerator(
         sensitiveLayers: value.sensitiveLayers.map((item) => typeof item === "string" ? item : item.layer),
         referenceIds: [],
         domainReasons: [],
-        evidenceRequest: request.phase === "final" ? null : {
+        evidenceRequest: request.phase === "final" || freeNarrative ? null : {
           domains,
           datePrecision: "month_preferred",
           prompt: unresolved?.dateLabel === "日期待补充"
@@ -224,6 +230,7 @@ function harness(options: {
   readonly detailQuestion?: string;
   readonly mislabelLatestDetailAsNewEvent?: boolean;
   readonly classifyEvidenceDomain?: (text: string) => "career" | "education" | "finance" | "health_pressure" | "relocation" | "relationship" | "family" | "other" | null;
+  readonly freeNarrativeFromGeneration?: number;
 } = {}) {
   const events: string[] = [];
   const narrativePrompts: string[] = [];
@@ -465,6 +472,7 @@ function harness(options: {
       options.detailQuestion,
       options.mislabelLatestDetailAsNewEvent,
       options.classifyEvidenceDomain,
+      options.freeNarrativeFromGeneration,
     ),
     asOfDate: () => "2026-07-21",
   };
@@ -525,14 +533,15 @@ test("start creates a deterministic opening without scanning or narrative genera
   assert.equal(turn.candidate.rangeEnd, "05:50");
   assert.match(turn.narrative, /当前先核对 04:50–05:50/);
   assert.match(turn.narrative, /还不能把其中某一分钟当作已确认出生时间/);
-  assert.match(turn.narrative, /请先说一件/);
+  assert.match(turn.narrative, /按自己的节奏/);
+  assert.match(turn.narrative, /一次说一件或连续说多件/);
   assert.deepEqual(turn.technicalReceipt, {
     calculationVersion: "rectification-opening-v1",
     stableLayers: [],
     sensitiveLayers: [],
     candidateDifferenceRefs: [],
   });
-  assert.equal(turn.evidenceRequest?.followUp?.kind, "new_event");
+  assert.equal(turn.evidenceRequest, null);
   assert.equal(JSON.stringify(turn).includes("candidateWeights"), false);
   assert.equal(value.cases.get(startActionId)?.row.revisionOfCaseId, priorCaseId);
   assert.equal(value.cases.get(startActionId)?.row.baselineActiveTime, "04:58");
@@ -1904,6 +1913,26 @@ test("vague, future, and unmatched answers stay conversational and never score",
   }
 });
 
+test("a free Agent reply clears the previous structured question instead of inheriting it", async () => {
+  const value = harness({ readyAfterEvidenceCount: 99, freeNarrativeFromGeneration: 2 });
+  await start(value, null);
+
+  const first = await value.service.answer(userId, {
+    type: "answer", caseId: startActionId, actionId: answerActionId,
+    turnVersion: 0, answer: "2020年4月进入研究院实习",
+  });
+  assert.ok(first.evidenceRequest);
+
+  const second = await value.service.answer(userId, {
+    type: "answer", caseId: startActionId, actionId: secondAnswerActionId,
+    turnVersion: 1, answer: "2020年10月主动辞职，之后准备考研",
+  });
+
+  assert.equal(second.evidenceRequest, null);
+  assert.match(second.narrative, /按自己的节奏/);
+  assert.doesNotMatch(second.narrative, /[？?]/);
+});
+
 test("a non-scoring packet failure responds to the current turn instead of replaying the prior agent message", async () => {
   const value = harness({
     packetFailure: new Error("synthetic packet outage"),
@@ -1917,8 +1946,8 @@ test("a non-scoring packet failure responds to the current turn instead of repla
   });
 
   assert.notEqual(turn.narrative, initial.narrative);
-  assert.match(turn.narrative, /化学专业|这轮|这次分析/);
-  assert.match(turn.narrative, /暂时没有完成/);
+  assert.match(turn.narrative, /化学专业|这轮|内容已经保留/);
+  assert.match(turn.narrative, /按自己的节奏/);
   assert.equal(value.cases.get(startActionId)?.row.validationReceipts.at(-1)?.fallbackUsed, true);
   assert.equal(value.cases.get(startActionId)?.row.eventEvidence.at(-1)?.rawText, "化学专业");
 });