diff --git a/BLOCKED.md b/BLOCKED.md
index 831cf41c..1ab9a618 100644
--- a/BLOCKED.md
+++ b/BLOCKED.md
@@ -8,6 +8,11 @@
 - **恢复演练**：内测开放前按 `docs/operations/production-restore-drill-20261014.md` 做一次一次性库还原，结果填 `docs/testing/production-restore-drill-20261014.md`。模板现在是空的。
 - 解除：上面四件事做完，并把演练模板里的「是否通过」填上。不要把 Webhook、密钥或用户内容写进仓库。
 
+## TASK-answer-effort-ab-20261013：答题步推理强度的真实 A/B（2026-10-15）
+
+- **A/B 未跑：缺模型 key。** 本机没有 `DS_KEY` 或 `BACKTEST_MODEL_KEY`。供应商是否接受 `low` / `medium` / `high` / `max`、推理 token 和耗时有没有下降，都没测。不得把请求体拦截写成「A/B 通过」。
+- 开关默认不设，请求体与改前一致。默认值等有 key 的验收跑完再由产品拍板，本单不改部署。
+
 ## TASK-consult-prep-parallel-20261013：审核真实延迟、Postgres 中位数、多领域镜像（2026-10-14）
 
 - **阿里云审核**：本机没有审核密钥。1500 毫秒挡首帧的结论来自测试里的假等待，不是真实供应商。开关 `CONSULT_PREP_PARALLEL` 保持关闭。
diff --git a/CHANGELOG.md b/CHANGELOG.md
index e3c2f28a..3e0cad6f 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -2,7 +2,11 @@
 
 ## 2026-10-15 — 内测前补上每日备份、宕机告警和日志上限（待合入 staging）
 
-- 生产可以每天自动做一次加密备份。站点连续三次打不开、磁盘过满，或短时间错误变多时，可以通知到飞书群。容器日志有大小上限。可以按一轮请求的编号在服务器日志里查找。这些要等产品在主机上装好之后才生效。页面没有变化。Skill 版本不变（BUG-1354、BUG-1355、BUG-1356）。
+- 生产可以每天自动做一次加密备份。站点连续三次打不开、磁盘过满，或短时间错误变多时，可以通知到飞书群。容器日志有大小上限。可以按一轮请求的编号在服务器日志里查找。这些要等产品在主机上装好之后才生效。页面没有变化。Skill 版本不变（BUG-1355、BUG-1356、BUG-1357）。
+
+## 2026-10-15 — 合盘更快（待合入 staging）
+
+- 合盘排两张盘时关掉用不上的外部概览，两张盘一起排。个人报告在没装参照引擎时不再为注定失败的回放反复起进程。轻重任务要不要分开限流，这一单只给了模拟表，还没改规则。
 
 ## 2026-10-15 — 星盘页上升行补上星宿，表头对齐（待合入 staging）
 
diff --git a/docs/BUG_HISTORY.md b/docs/BUG_HISTORY.md
index 481957cd..1c135ef4 100644
--- a/docs/BUG_HISTORY.md
+++ b/docs/BUG_HISTORY.md
@@ -15304,17 +15304,17 @@
 ## BUG-1140 | staging 镜像打包失败：`frontend/scripts` 下的研究脚本引用了 `frontend/tests` 的夹具
 
 - 状态：resolved（修复已推 staging；以门禁 publish 通过与 `/api/health` 部署版本为最终证据）
-- 首次发现 / 最近更新：2026-10-01 / 2026-10-14
+- 首次发现 / 最近更新：2026-10-01 / 2026-10-15
 - 影响面：Gitea `backend-quality-gate` 的 publish 作业（`deploy/railway-web.Dockerfile` 构建阶段 `npm run build`）。validate 通过、publish 失败，未部署，staging 停在上一版。
 - 用户现象：盘型口径（BUG-1115～1117、1131）推送后 staging 不更新。
 - 触发条件：`frontend/scripts/rectification-segment-persisted-replay.ts`（盘型口径研究的 PG 持久回放 harness）`import … from "../tests/helpers/postgres-fixture.ts"`。web 镜像只复制 `frontend/src`、`frontend/scripts` 等，不复制 `frontend/tests`；`frontend/tsconfig.json` 的 include 为 `**/*.ts`，`next build` 类型检查覆盖 scripts，于是 `TS2307 Cannot find module '../tests/helpers/postgres-fixture.ts'`。
 - 根因：本地与验收机都有 `frontend/tests`，`tsc --noEmit` 与 `next build` 均通过，门禁 validate 也在完整检出上跑，没有任何一步按镜像的复制清单构建。
 - 修复：harness 移到 `frontend/tests/research/rectification-segment-persisted-replay.ts`（相对路径随之调整），Python driver `scripts/research/varga_resolution_persisted_replay.py` 指向新路径；新增 `frontend/tests/runtime-sources-no-test-imports.test.ts`：`src/` 与 `scripts/` 不得 import / `new URL` 引用 `tests/`。
 - 验证：按 `railway-web.Dockerfile` 的 COPY 清单从提交 `git archive` 出构建目录，修复前 `npm run build` 报上述 TS2307、exit 1，修复后 exit 0、`/` Static；守卫测试 2/2，且对旧 harness 文本判定为违规；移动后 harness 1 例 ±10 冒烟 `runner_exit=0`。
-- 防复发：守卫测试进 `npm test`（门禁 validate 必跑）；错误台账 ERR-113。2026-10-14 这道守卫在合入前拦住了同一形状：issue #58 的往返对照脚本曾放在 `frontend/scripts/` 并导入测试夹具。脚本改放到 `frontend/tests/bench-local-pg-roundtrips.ts`，文件名不是 `.test.ts`，不会被全量测试收进去。业务代码没有改。原任务书里的旧路径保持不动。
+- 防复发：守卫测试进 `npm test`（门禁 validate 必跑）；错误台账 ERR-113。2026-10-14 这道守卫在合入前拦住了同一形状：issue #58 的往返对照脚本曾放在 `frontend/scripts/` 并导入测试夹具。脚本改放到 `frontend/tests/bench-local-pg-roundtrips.ts`，文件名不是 `.test.ts`，不会被全量测试收进去。业务代码没有改。原任务书里的旧路径保持不动。2026-10-15 issue #65 的请求体探针也是同一形状：`answer-effort-probe.mts` 放在 `frontend/scripts/research/`，用 `new URL` 去读 `frontend/tests` 的夹具。探针挪到 `frontend/tests/answer-effort-probe.mts`，文件名不是 `.test.ts`。业务代码没有改。
 - 相关记录：BUG-1115～1117、BUG-1131、BUG-1327。
-- 复发自：无。2026-10-14 是同一形状，被守卫拦住，未进测试环境。
-- 修复版本：`codex/staging-publish-fix-20261001`。2026-10-14 的同类挪动在 `codex/node-runtime-resources-20261013`。
+- 复发自：无。2026-10-14 与 2026-10-15 都是同一形状，被守卫拦住，未进测试环境。
+- 修复版本：`codex/staging-publish-fix-20261001`。2026-10-14 的同类挪动在 `codex/node-runtime-resources-20261013`。2026-10-15 的同类挪动在 `codex/answer-effort-ab-20261013`。
 
 ## BUG-1141 | 生时校正缺少度数级时间信号（角度类技法研究）
 
@@ -17991,6 +17991,36 @@
 - 复发自：无。
 - 修复版本：分支 `codex/api-disk-cache-retention-20261013`，尚未合入 staging。
 
+## BUG-1333 | 合盘排盘漏关外部概览，而且两张盘排队算
+
+- 状态：fixed-pending-verify
+- 首次发现 / 最近更新：2026-10-13 / 2026-10-15
+- 来源：issue #64。任务书 `docs/tasks/TASK-api-chart-subprocess-gate-20261013.md` T1、T3、T4。
+- 现象：合盘每次要排两张盘，都没关掉用不上的外部概览。每张盘多起两个后台进程，而且两张盘一张一张排。名额占满时，日志里看不出是谁占着。
+- 触发条件：打开合盘。排盘接口本身默认仍走完整链，给报告和外部调用留着。
+- 根因：合盘这条路由没跟上「只读盘面就关掉外部概览」的约定，也没有合同测试拦住。名额按路径一刀切，不带 skip 的排盘默认不占名额。
+- 修复：合盘两次排盘都带上关闭外部概览，两张盘同时排；婚恋分支的两张分盘也同时排。前端里凡是发给排盘接口的请求体，都要带这个开关，由源码合同守着。另加一个默认关闭的开关：打开后，不带 skip 的同步排盘才占一个重计算名额。名额占满时记一行日志，只写被拒的路径和当前占用者的路径、已占秒数，不写请求正文。
+- 验证：`frontend/tests/synastry-chart-skip.test.ts`、`frontend/tests/chart-request-skip-contract.test.ts`、`tests/test_api_chart_subprocess_gate.py`、`tests/test_api_heavy_compute_gate.py`。合盘耗时见 `docs/testing/api-chart-subprocess-gate-ab-20261013.md`。`tests/test_supabase_user_data_contract.py` 锁住合盘路由向 `/api/chart` 请求本人和对方两张盘，且两次都带 `skip_vedastro_main_entry_overview: true`。
+- 防复发：源码合同要求 `frontend/src/app/api` 和 `frontend/src/lib` 里发往 `/api/chart` 的请求体带 `skip_vedastro_main_entry_overview: true`。Python 合同按路由原文锁住这两次调用和这个开关。`/api/chart` 仍不在重计算路径名单里。
+- 相关记录：BUG-161、BUG-715、BUG-717、BUG-1302。
+- 复发自：无。
+- 修复版本：分支 `codex/api-chart-subprocess-gate-20261013`。编号是任务书预留的空号，夹在 BUG-1332 和 BUG-1335 之间。
+
+## BUG-1334 | 个人报告在没有参照引擎时仍为每次报告启动子进程
+
+- 状态：fixed-pending-verify
+- 首次发现 / 最近更新：2026-10-13 / 2026-10-15
+- 来源：issue #64。任务书 T2。
+- 现象：每份个人报告会起大约 10 个后台进程。其中参照引擎没装时，年度回放注定失败，但仍要付启动开销。另一段静态校验和用户的盘无关，每份报告都重跑。
+- 触发条件：生成个人长报告，且这台机器没有安装 PyJHora。装了参照引擎的开发机上，隔离用的子进程仍然要留着。
+- 根因：子进程是为了把参照引擎隔离开，这是对的；父进程却没有先问「这台机器根本没装」。静态校验文件被挂在每份报告上，没有按文件变化复用。
+- 修复：父进程先看一次能不能导入参照引擎，不能就不起子进程，返回和「子进程在没装时」相同的结果。解释器查找和静态校验各在进程内记一次；校验文件变了会重算。不缓存任何出生资料。装了参照引擎时仍走子进程。
+- 验证：`tests/test_api_chart_subprocess_gate.py` 对照 `tests/fixtures/api-chart-subprocess-gate/`。生产上的收益大约是每份报告 0.5 到 0.9 秒、少掉那几个注定失败的进程，不是排查时在装了参照引擎的机器上看到的 5 秒。本机没有 PyJHora，实测见 `docs/testing/api-chart-subprocess-gate-ab-20261013.md`。
+- 防复发：无参照引擎时测试断言不会调用 `subprocess.run`，返回值与固定样本一致。
+- 相关记录：BUG-721、BUG-1265、BUG-1333。
+- 复发自：无。
+- 修复版本：分支 `codex/api-chart-subprocess-gate-20261013`。
+
 ## BUG-1335 | 个人长报告的中文出稿和溯源哈希重复劳动
 
 - 状态：investigating
@@ -33372,17 +33402,17 @@
 ## BUG-1140 | staging 镜像打包失败：`frontend/scripts` 下的研究脚本引用了 `frontend/tests` 的夹具
 
 - 状态：resolved（修复已推 staging；以门禁 publish 通过与 `/api/health` 部署版本为最终证据）
-- 首次发现 / 最近更新：2026-10-01 / 2026-10-14
+- 首次发现 / 最近更新：2026-10-01 / 2026-10-15
 - 影响面：Gitea `backend-quality-gate` 的 publish 作业（`deploy/railway-web.Dockerfile` 构建阶段 `npm run build`）。validate 通过、publish 失败，未部署，staging 停在上一版。
 - 用户现象：盘型口径（BUG-1115～1117、1131）推送后 staging 不更新。
 - 触发条件：`frontend/scripts/rectification-segment-persisted-replay.ts`（盘型口径研究的 PG 持久回放 harness）`import … from "../tests/helpers/postgres-fixture.ts"`。web 镜像只复制 `frontend/src`、`frontend/scripts` 等，不复制 `frontend/tests`；`frontend/tsconfig.json` 的 include 为 `**/*.ts`，`next build` 类型检查覆盖 scripts，于是 `TS2307 Cannot find module '../tests/helpers/postgres-fixture.ts'`。
 - 根因：本地与验收机都有 `frontend/tests`，`tsc --noEmit` 与 `next build` 均通过，门禁 validate 也在完整检出上跑，没有任何一步按镜像的复制清单构建。
 - 修复：harness 移到 `frontend/tests/research/rectification-segment-persisted-replay.ts`（相对路径随之调整），Python driver `scripts/research/varga_resolution_persisted_replay.py` 指向新路径；新增 `frontend/tests/runtime-sources-no-test-imports.test.ts`：`src/` 与 `scripts/` 不得 import / `new URL` 引用 `tests/`。
 - 验证：按 `railway-web.Dockerfile` 的 COPY 清单从提交 `git archive` 出构建目录，修复前 `npm run build` 报上述 TS2307、exit 1，修复后 exit 0、`/` Static；守卫测试 2/2，且对旧 harness 文本判定为违规；移动后 harness 1 例 ±10 冒烟 `runner_exit=0`。
-- 防复发：守卫测试进 `npm test`（门禁 validate 必跑）；错误台账 ERR-113。2026-10-14 这道守卫在合入前拦住了同一形状：issue #58 的往返对照脚本曾放在 `frontend/scripts/` 并导入测试夹具。脚本改放到 `frontend/tests/bench-local-pg-roundtrips.ts`，文件名不是 `.test.ts`，不会被全量测试收进去。业务代码没有改。原任务书里的旧路径保持不动。
+- 防复发：守卫测试进 `npm test`（门禁 validate 必跑）；错误台账 ERR-113。2026-10-14 这道守卫在合入前拦住了同一形状：issue #58 的往返对照脚本曾放在 `frontend/scripts/` 并导入测试夹具。脚本改放到 `frontend/tests/bench-local-pg-roundtrips.ts`，文件名不是 `.test.ts`，不会被全量测试收进去。业务代码没有改。原任务书里的旧路径保持不动。2026-10-15 issue #65 的请求体探针也是同一形状：`answer-effort-probe.mts` 放在 `frontend/scripts/research/`，用 `new URL` 去读 `frontend/tests` 的夹具。探针挪到 `frontend/tests/answer-effort-probe.mts`，文件名不是 `.test.ts`。业务代码没有改。
 - 相关记录：BUG-1115～1117、BUG-1131、BUG-1327。
-- 复发自：无。2026-10-14 是同一形状，被守卫拦住，未进测试环境。
-- 修复版本：`codex/staging-publish-fix-20261001`。2026-10-14 的同类挪动在 `codex/node-runtime-resources-20261013`。
+- 复发自：无。2026-10-14 与 2026-10-15 都是同一形状，被守卫拦住，未进测试环境。
+- 修复版本：`codex/staging-publish-fix-20261001`。2026-10-14 的同类挪动在 `codex/node-runtime-resources-20261013`。2026-10-15 的同类挪动在 `codex/answer-effort-ab-20261013`。
 
 ## BUG-1141 | 生时校正缺少度数级时间信号（角度类技法研究）
 
@@ -36044,6 +36074,36 @@
 - 复发自：无。
 - 修复版本：分支 `codex/api-disk-cache-retention-20261013`，尚未合入 staging。
 
+## BUG-1333 | 合盘排盘漏关外部概览，而且两张盘排队算
+
+- 状态：fixed-pending-verify
+- 首次发现 / 最近更新：2026-10-13 / 2026-10-15
+- 来源：issue #64。任务书 `docs/tasks/TASK-api-chart-subprocess-gate-20261013.md` T1、T3、T4。
+- 现象：合盘每次要排两张盘，都没关掉用不上的外部概览。每张盘多起两个后台进程，而且两张盘一张一张排。名额占满时，日志里看不出是谁占着。
+- 触发条件：打开合盘。排盘接口本身默认仍走完整链，给报告和外部调用留着。
+- 根因：合盘这条路由没跟上「只读盘面就关掉外部概览」的约定，也没有合同测试拦住。名额按路径一刀切，不带 skip 的排盘默认不占名额。
+- 修复：合盘两次排盘都带上关闭外部概览，两张盘同时排；婚恋分支的两张分盘也同时排。前端里凡是发给排盘接口的请求体，都要带这个开关，由源码合同守着。另加一个默认关闭的开关：打开后，不带 skip 的同步排盘才占一个重计算名额。名额占满时记一行日志，只写被拒的路径和当前占用者的路径、已占秒数，不写请求正文。
+- 验证：`frontend/tests/synastry-chart-skip.test.ts`、`frontend/tests/chart-request-skip-contract.test.ts`、`tests/test_api_chart_subprocess_gate.py`、`tests/test_api_heavy_compute_gate.py`。合盘耗时见 `docs/testing/api-chart-subprocess-gate-ab-20261013.md`。`tests/test_supabase_user_data_contract.py` 锁住合盘路由向 `/api/chart` 请求本人和对方两张盘，且两次都带 `skip_vedastro_main_entry_overview: true`。
+- 防复发：源码合同要求 `frontend/src/app/api` 和 `frontend/src/lib` 里发往 `/api/chart` 的请求体带 `skip_vedastro_main_entry_overview: true`。Python 合同按路由原文锁住这两次调用和这个开关。`/api/chart` 仍不在重计算路径名单里。
+- 相关记录：BUG-161、BUG-715、BUG-717、BUG-1302。
+- 复发自：无。
+- 修复版本：分支 `codex/api-chart-subprocess-gate-20261013`。编号是任务书预留的空号，夹在 BUG-1332 和 BUG-1335 之间。
+
+## BUG-1334 | 个人报告在没有参照引擎时仍为每次报告启动子进程
+
+- 状态：fixed-pending-verify
+- 首次发现 / 最近更新：2026-10-13 / 2026-10-15
+- 来源：issue #64。任务书 T2。
+- 现象：每份个人报告会起大约 10 个后台进程。其中参照引擎没装时，年度回放注定失败，但仍要付启动开销。另一段静态校验和用户的盘无关，每份报告都重跑。
+- 触发条件：生成个人长报告，且这台机器没有安装 PyJHora。装了参照引擎的开发机上，隔离用的子进程仍然要留着。
+- 根因：子进程是为了把参照引擎隔离开，这是对的；父进程却没有先问「这台机器根本没装」。静态校验文件被挂在每份报告上，没有按文件变化复用。
+- 修复：父进程先看一次能不能导入参照引擎，不能就不起子进程，返回和「子进程在没装时」相同的结果。解释器查找和静态校验各在进程内记一次；校验文件变了会重算。不缓存任何出生资料。装了参照引擎时仍走子进程。
+- 验证：`tests/test_api_chart_subprocess_gate.py` 对照 `tests/fixtures/api-chart-subprocess-gate/`。生产上的收益大约是每份报告 0.5 到 0.9 秒、少掉那几个注定失败的进程，不是排查时在装了参照引擎的机器上看到的 5 秒。本机没有 PyJHora，实测见 `docs/testing/api-chart-subprocess-gate-ab-20261013.md`。
+- 防复发：无参照引擎时测试断言不会调用 `subprocess.run`，返回值与固定样本一致。
+- 相关记录：BUG-721、BUG-1265、BUG-1333。
+- 复发自：无。
+- 修复版本：分支 `codex/api-chart-subprocess-gate-20261013`。
+
 ## BUG-1335 | 个人长报告的中文出稿和溯源哈希重复劳动
 
 - 状态：investigating
@@ -36182,7 +36242,22 @@
 - 复发自：无
 - 修复版本：分支 `codex/chart-ascendant-row-20261015`。
 
-## BUG-1354 | 生产没有每日自动备份
+## BUG-1354 | 低强度推理设置写在嵌套字段里，供应商收不到
+
+- 状态：investigating
+- 首次发现 / 最近更新：2026-10-13 / 2026-10-15
+- 来源：issue #65。任务书 `docs/tasks/TASK-answer-effort-ab-20261013.md`。请求体拦截见 `docs/testing/answer-effort-ab-20261013.md`。
+- 现象：代码里有一处把推理强度写成 `low`，但真正发出去的请求没有 `reasoning_effort` 这个字段。强度被包在 `thinking` 里面。
+- 触发条件：调用 `consultationSliceGenerationSettings`。产品代码里没有调用方，线上对话不受影响。
+- 根因：兼容通道只认供应商选项最外层的 `reasoningEffort`，并把它写成 `reasoning_effort`。嵌在 `thinking` 里的同名键会原样送出去，供应商不把它当强度。
+- 修复：本单不改这处形状，也不改它的两条断言。答题步的新开关写在最外层。要不要删掉或改掉这处旧形状，等产品拍板。
+- 验证：`frontend/tests/answer-effort-probe.mts --dry-run`。这处设置的请求体是 `thinking.reasoningEffort = "low"`，`reasoning_effort` 为空。新开关设成 `low` 时，答题步请求体有 `reasoning_effort: "low"`。
+- 防复发：新测试锁住「开关不设时请求体与改前一致」，并锁住新强度出现在最外层。
+- 相关记录：BUG-148、BUG-1142。
+- 复发自：无。
+- 修复版本：未修。分支 `codex/answer-effort-ab-20261013` 只记录。原先写在 BUG-1351，该号已被数据层占用；按本文件规则保留较早的 BUG-1351，本条改为 BUG-1354，接在 BUG-1353 之后。
+
+## BUG-1355 | 生产没有每日自动备份
 
 - 状态：代码已交，主机安装待产品
 - 首次发现 / 最近更新：2026-10-14 / 2026-10-15
@@ -36193,37 +36268,37 @@
 - 修复：新增只从 `main` 运行的每日备份工作流，复用现有加密与还原校验脚本。每日文件用单独前缀，过期清理不会碰到手动恢复点。工作流要等进入 `main` 后才会在北京时间 04:00 跑。本机没有生产主机，没有手动触发。
 - 验证：合同测试锁定有每日时刻、只在 `main` 上跑、调用同一个恢复脚本、清理只删过期的每日前缀。虚构目录里，手动恢复点和其他文件都还在。
 - 防复发：上述合同测试。清理函数不接受外部传入的前缀。
-- 相关记录：BUG-1350、BUG-1355、BUG-1356。
+- 相关记录：BUG-1350、BUG-1356、BUG-1357。
 - 复发自：无。
-- 修复版本：分支 `codex/beta-safety-net-20261014`。开工时本文件最大号是 BUG-1353。
+- 修复版本：分支 `codex/beta-safety-net-20261014`。开工时本文件最大号是 BUG-1353，当时写成 BUG-1354。合并 staging 后该号已被更早的答题强度记录占用，按台账规则改为 BUG-1355。
 
-## BUG-1355 | 站点打不开或磁盘过满时没有告警
+## BUG-1356 | 站点打不开或磁盘过满时没有告警
 
 - 状态：代码已交，主机安装待产品
 - 首次发现 / 最近更新：2026-10-14 / 2026-10-15
-- 来源：同 BUG-1354。
+- 来源：同 BUG-1355。
 - 现象：容器自己的健康检查只会尝试重启，不会通知人。磁盘变满也没有通知。
 - 触发条件：公网健康检查连续失败，或生产主机磁盘使用率达到 80%。
 - 根因：上线时没有持续探活，也没有把告警发到人用的渠道。
 - 修复：探活脚本连续 3 次失败才发一条告警，恢复只发一次。磁盘检查单独在生产主机上跑。Webhook 从环境文件读取，没有配置就只写本机日志。外发正文不含用户内容。脚本还没装到 runner 或生产主机上。
 - 验证：本地桩服务覆盖正常、500、超时和缺少字段。连续 3 次才告警，恢复只出现一次。签名测试里的密钥没有出现在发出的正文里。
 - 防复发：上述脚本测试。告警行若含不允许的字段，脚本拒绝发送。
-- 相关记录：BUG-1354、BUG-1356。
+- 相关记录：BUG-1355、BUG-1357。
 - 复发自：无。
-- 修复版本：分支 `codex/beta-safety-net-20261014`。编号接在 BUG-1354 之后。
+- 修复版本：分支 `codex/beta-safety-net-20261014`。原先写成 BUG-1355，合并 staging 后顺延为 BUG-1356。
 
-## BUG-1356 | 容器日志不轮转，也不能按一轮请求查找
+## BUG-1357 | 容器日志不轮转，也不能按一轮请求查找
 
 - 状态：代码已交，主机安装待产品
 - 首次发现 / 最近更新：2026-10-14 / 2026-10-15
-- 来源：同 BUG-1354。BUG-1306 取证时日志只在容器控制台，重建后查不到。
+- 来源：同 BUG-1355。BUG-1306 取证时日志只在容器控制台，重建后查不到。
 - 现象：编排没有日志大小上限。知道一轮请求的编号时，没有只读的查找命令。
 - 触发条件：日志持续写入，或需要按 requestId 回看 web、api、caddy。
 - 根因：日志沿用守护进程的未知默认值，也没有按编号检索的入口。
 - 修复：每个服务的日志改为单文件 20 MB、最多 10 个文件。新增只读查找脚本，只接受 UUID。错误汇总只计数和最多 5 个编号，丢掉含邮箱、额外日期或非 ASCII 文本的行。要等生产容器按新编排重建后，轮转才生效。
 - 验证：两个编排文件里每个服务的日志段一致。虚构日志能按时间排序找到编号；命令注入写法被拒绝，旁边的标记文件还在。含邮箱和汉字的样本没有进入告警文本。
 - 防复发：上述合同测试和查找脚本的格式检查。
-- 相关记录：BUG-1306、BUG-1354、BUG-1355。
+- 相关记录：BUG-1306、BUG-1355、BUG-1356。
 - 复发自：无。
-- 修复版本：分支 `codex/beta-safety-net-20261014`。编号接在 BUG-1355 之后。
+- 修复版本：分支 `codex/beta-safety-net-20261014`。原先写成 BUG-1356，合并 staging 后顺延为 BUG-1357。
 
diff --git a/docs/tasks/PROGRESS-answer-effort-ab-20261013.md b/docs/tasks/PROGRESS-answer-effort-ab-20261013.md
new file mode 100644
index 00000000..b93839fe
--- /dev/null
+++ b/docs/tasks/PROGRESS-answer-effort-ab-20261013.md
@@ -0,0 +1,43 @@
+# PROGRESS · 答题步推理强度开关（2026-10-15）
+
+- 分支：`codex/answer-effort-ab-20261013`
+- 实际基线：`origin/staging` `9274491681f6fcacbc87e3797441d2c5221de874`
+- 前置 `consult-prep-parallel` 已在 `origin/staging` 上（`git merge-base --is-ancestor` 退出码 0）
+- 默认行为不变。没有模型钥匙，A/B 未跑。
+
+## 做了什么
+
+1. 新环境变量 `CONSULT_ANSWER_REASONING_EFFORT`。`low` / `medium` / `high` / `max` 才生效。空着或没设，请求体与改前一致。非法值 `console.warn` 一次，然后当作没设。
+2. `agentGenerationSettings` 增加 `providerReasoningEffort`。它写在供应商选项的最外层，兼容通道据此发出 `reasoning_effort`。原来的 `reasoningEffort` 仍嵌在 `thinking` 里，切片那两条断言不用改。
+3. 答题步才带这个字段。判据与答案时钟相同：计算结果已经在手（`answerReady`），或者公开日历从第一步就是答题（`answerFromStart`）。主循环第 0 步计算结果还没到，不带。重试是新的一轮，步号从 0 重新计，但计算结果已经在手，所以重试的第 0 步也带。续写仍是关闭思考，不带这个字段。
+4. Mastra 的 `prepareStep` 会按步合并 `providerOptions`（`mergeObjects`）。答题步返回的强度会并进这一步，不会整轮都带。本单按步生效，没有改成整轮。
+5. 生平回测脚本增加 `--cases figure:domain,…` 和 `--efforts unset,low,medium`。每行记下端到端秒数、第 0 / 1 步秒数和推理 token、输出 token、结束原因、是否空答、是否截断、正文字数。摘要记下当次强度。
+
+## 自测
+
+| 检查 | 命令 | 结果 |
+| --- | --- | --- |
+| 请求体拦截 | `node node_modules/tsx/dist/cli.mjs tests/answer-effort-probe.mts --dry-run` | 退出码 0。数字见 `docs/testing/answer-effort-ab-20261013.md`。2026-10-15 打回后脚本从 `frontend/scripts/research/` 挪到 `frontend/tests/`，命令改成这一条。 |
+| 本单测试与任务书点名的四份 | `node node_modules/tsx/dist/cli.mjs --test` 后接 `answer-effort-switch-20261013`、`consultation-agentic-runtime`、`consultation-workflow-contract`、`consult-answer-clock-20261001`、`consult-latency-timing-20261005` | 114 pass，0 fail，0 cancelled |
+| 类型检查 | 在 `frontend` 下 `.\node_modules\.bin\tsc --noEmit -p .` | 退出码 0 |
+| lint | 在 `frontend` 下 `npm run lint` | 退出码 0，0 error，123 条原有 warning。探针里多出来的 1 条未使用变量已去掉，改过的文件再跑 eslint 是 0 |
+
+开关不设时，两步请求体与改前保存的整份 JSON 逐字节一致。设 `low` 时只有答题步出现 `reasoning_effort: "low"`，第 0 步没有；续写仍是 `thinking: disabled` 且没有 `reasoning_effort`。
+
+## 改既有断言
+
+| 原值 | 新值 | 原因 |
+| --- | --- | --- |
+| `await streamWithOverflowRetry(agent)` | `await streamWithOverflowRetry(agent, generalStreamOptions)` | 公开日历从第一步就是答题。不传选项就没有 `prepareStep`，强度加不上去。开关不设时这些选项不改请求体。答案时钟的断言还在。 |
+
+切片的嵌套形状和它的两条断言没有改。
+
+## 需拍板
+
+- 默认强度。本单不改。有钥匙的验收按任务书 T3 出数之后，再决定设不设、设哪一档。部署默认写在哪里也留到那次。
+- `consultationSliceGenerationSettings` 删掉还是改成最外层。它没有调用方。本单不改。见 BUG-1354（原先写成 BUG-1351，该号已被数据层占用）。
+- 若供应商不接受降档，对照组 D（答题步关思考）只供参考，采用与否另说。本机没跑。
+
+## 环境缺口
+
+A/B 未跑：缺模型 key。没有 `DS_KEY`，也没有 `BACKTEST_MODEL_KEY`。
diff --git a/docs/tasks/PROGRESS-answer-effort-ab-fix-20261015.md b/docs/tasks/PROGRESS-answer-effort-ab-fix-20261015.md
new file mode 100644
index 00000000..b6a48f5b
--- /dev/null
+++ b/docs/tasks/PROGRESS-answer-effort-ab-fix-20261015.md
@@ -0,0 +1,40 @@
+# PROGRESS · 答题推理强度开关修复（2026-10-15）
+
+- 分支：`codex/answer-effort-ab-20261013`
+- 被打回的提交：`ab7f66be14d93a5ccbc3433f3c44e134a65b1e21`
+- 修复单：`docs/tasks/TASK-answer-effort-ab-fix-20261015.md`
+- 已合并 `origin/staging` `71f7fac3a13f817f91755ff01b25b8706380ae19`，合并提交 `0d73cd332bf4efae2b8ec9389d5853384f4616f4`。没有改写已推送的历史。
+
+## 做了什么
+
+1. 把 `frontend/scripts/research/answer-effort-probe.mts` 挪到 `frontend/tests/answer-effort-probe.mts`。文件名不是 `.test.ts`，也不是 `.test.tsx`，全量测试不会把它收成一条用例。
+2. 相对路径改过：源码从 `../src/...` 引入，夹具改成 `./fixtures/answer-effort-switch-20261013.json`。本单测试改为从同目录引入。已经保存的夹具文件没有重写。
+3. 命令同步进 `docs/tasks/PROGRESS-answer-effort-ab-20261013.md` 和 `docs/testing/answer-effort-ab-20261013.md`。
+4. 业务代码没改。开关默认值没改。任务书里留到拍板的默认强度，以及没人调用的旧嵌套设置，都没做。
+5. 合并时，本分支原来的调查记录占用了 BUG-1351，和数据层的 BUG-1351 撞号。保留 staging 上的 BUG-1351、BUG-1352、BUG-1353。这条调查记录改为 BUG-1354。BUG-1140 补记了这次同一形状的挪动。
+
+## 自测
+
+| 检查 | 命令 | 结果 |
+| --- | --- | --- |
+| 镜像守卫和本单测试 | 在 `frontend` 下 `node node_modules/tsx/dist/cli.mjs --test tests/runtime-sources-no-test-imports.test.ts tests/answer-effort-switch-20261013.test.ts` | 5 pass，0 fail，0 cancelled |
+| 请求体拦截 | 在 `frontend` 下 `node node_modules/tsx/dist/cli.mjs tests/answer-effort-probe.mts --dry-run` | 退出码 0 |
+| 类型检查 | 在 `frontend` 下 `.\node_modules\.bin\tsc --noEmit -p .` | 退出码 0 |
+| lint | 在 `frontend` 下 `npm run lint` | 退出码 0，0 error，123 warning |
+| 隐私标记与 bug 记录格式 | 在工作树根 `.venv\Scripts\python.exe -m pytest tests/test_repo_privacy_markers.py tests/test_bug_history_workflow.py -q` | 64 pass，0 fail，退出码 0 |
+
+守卫的两条都在上面 5 条里面：镜像目录不再引用测试目录，并且仍认得当初弄坏构建的那几种写法。
+
+## prepareStep 会不会盖掉原有选项
+
+不会整份换掉。开关打开时，答题步返回的供应商选项来自 `agentGenerationSettings`，Mastra 按步做深合并，并进这一步已有的选项。思考仍是打开。输出长度上限在基础流的模型设置里，这一步没有返回它，所以上限还在。缓存标记写在消息上，不在这份流选项里。因此没有往 `BLOCKED.md` 新写一条「被覆盖」。
+
+## 偏离
+
+- 没有跑前端全量 `npm test`。这一侧的约定是不跑全量，留给验收。本机只跑了打回指出的那条守卫，以及本单测试。
+- 修复单写的 `git diff ab7f66be..<新提交> -- frontend/src` 不是空的。原因是按修复单先合并了最新 staging，staging 自己改过星盘页和数据层。本轮修复相对合并提交 `0d73cd33`，`frontend/src` 没有差异。业务代码和开关默认值没动。
+- A/B 未跑：缺模型 key。没有把请求体拦截写成对照已通过。
+
+## 环境缺口
+
+A/B 未跑：缺模型 key。`BLOCKED.md` 里原有的这一条保留，没有新增。
diff --git a/docs/tasks/PROGRESS-api-chart-subprocess-gate-20261013.md b/docs/tasks/PROGRESS-api-chart-subprocess-gate-20261013.md
new file mode 100644
index 00000000..93f936a1
--- /dev/null
+++ b/docs/tasks/PROGRESS-api-chart-subprocess-gate-20261013.md
@@ -0,0 +1,90 @@
+# PROGRESS · 合盘排盘与报告多余子进程（2026-10-15）
+
+- 分支：`codex/api-chart-subprocess-gate-20261013`
+- 实际基线：`origin/staging` `9274491681f6fcacbc87e3797441d2c5221de874`（任务书写的 `77057a819` 更早）
+- `scripts/jyotish_api_server.py`：开工 10944 行，做完 10940 行
+- BUG-1333、BUG-1334。`docs/BUG_HISTORY.md` 后半是同一份历史的副本，两处都补上了。
+
+## 做了什么
+
+1. 合盘两次排盘都带 `skip_vedastro_main_entry_overview: true`，不改 `synastryBirthPayload`。两张盘和婚恋分支的两张九分盘都改成同时发出。任一失败仍走原来的 `catch`。
+2. 前端 `app/api` 和 `lib` 里发往 `/api/chart` 的请求体必须带这个开关。白名单是空的。
+3. 父进程先看一次能不能导入 PyJHora。不能就不起年度回放和 Tajika 的子进程，返回和「子进程在没装时」相同的结果。样本在 `tests/fixtures/api-chart-subprocess-gate/`，来自本机真实子进程（当时 `find_spec("jhora")` 为 `None`）。装了之后仍走子进程。
+4. 解释器查找（含「没找到」）和静态 oracle 校验在进程内各记一次。校验按文件路径、mtime、大小作键。不缓存出生资料。
+5. 开关 `JYOTISH_GATE_CHART_OVERVIEW` 默认关。只有把它设为 `1` 时，不带 skip、也不是异步排队的 `/api/chart` 才占一个重计算名额。`/api/chart` 没有加进 `HEAVY_COMPUTE_PATHS`。预热咨询不占名额的判断挪进了闸门模块。
+6. 名额占满时打一行日志：被拒路径、当前占用者路径、已占秒数。不写请求正文。
+
+## 自测
+
+| 检查 | 命令 | 结果 |
+| --- | --- | --- |
+| 闸门、增长护栏、本单新测试 | `.venv\Scripts\python.exe -m pytest tests/test_api_chart_subprocess_gate.py tests/test_api_heavy_compute_gate.py tests/test_api_server_growth_contract.py -q` | 21 passed，退出码 0 |
+| 报告三份既有测试 | 同上解释器 `-m pytest tests/test_report_reader_main.py tests/test_report_longform_parity.py tests/test_report_density_facts.py -q` | 27 passed，退出码 0 |
+| 前端本单与相关单文件 | 在 `frontend` 下 `node node_modules/tsx/dist/cli.mjs --test tests/synastry-chart-skip.test.ts tests/chart-request-skip-contract.test.ts tests/chart-view-route.test.ts tests/ephemeris-route.test.ts` | 22 pass，0 fail |
+| 类型检查 | `frontend\node_modules\.bin\tsc --noEmit -p frontend` | 退出码 0 |
+| lint | 在 `frontend` 下 `npm run lint` | 退出码 0，0 error，123 条原有 warning |
+| A/B | `.venv\Scripts\python.exe scripts\research\api_chart_subprocess_ab.py` | 退出码 0。数字见下 |
+
+`grep` 这三个名字在 `tests/` 里只命中本单新测试：`annual_pyjhora_replay`、`_oracle_progress_snapshot`、`_resolve_external_pyjhora_python`。
+
+## A/B 数字
+
+合盘：串行不带 skip 中位 8.0596 秒、2 次 `subprocess.run`；并发带 skip 中位 0.0166 秒、0 次。盘面摘要逐字节一致。耗时约为对照的 0.2%。
+
+报告：强制仍起子进程中位 16.5322 秒、7 次；短路后 15.3235 秒、1 次。去掉易变字段后逐字节一致。少掉的 6 次是年度回放和 Tajika。
+
+名额模拟（种子 64，一分钟，未改规则）：
+
+| 方案 | 咨询被拒 | 报告被拒 | 校正被拒 | 最高并发 | 碰到不得加池的那条 |
+| --- | --- | --- | --- | --- | --- |
+| O1 现状 | 40% | 0% | 17% | 2 | 否 |
+| O2 咨询单独小池 | 0% | 0% | 17% | 3 | 是 |
+| O3 咨询最多再等 1.5 秒 | 30% | 0% | 33% | 2 | 否 |
+| O4 重任务最多占 1 个 | 0% | 0% | 67% | 2 | 否 |
+
+全文在 `docs/testing/api-chart-subprocess-gate-ab-20261013.md`。
+
+## 改既有断言
+
+无。原值 / 新值 / 原因：没有改旧断言。
+
+## 偏离
+
+- `/api/health` 没有加 `heavy_compute` 字段。状态函数 `heavy_compute_status()` 在闸门模块里，接到健康检查会让主文件变长。任务书允许砍掉这一项。
+- 星盘名额开关默认关。打开会让不带 skip 的外部调用可能拿到 429，任务书要求先有 A/B 再提议，本单不打开。
+- O2、O3、O4 没有实现。
+- 报告对照不是 staging 源码另导入，而是本分支把「已安装」强制打开。静态校验缓存两边都在，所以耗时差只反映那 6 个子进程。
+- 合盘子进程是 2 → 0，不是任务书举例的 4 → 0。这台机器两张盘合计只数到 2 次 `subprocess.run`。
+- 没有「装了 PyJHora」的计时。父进程强制当成已安装时，子进程仍会因为没装而失败，这一组用来代表「不短路就要付启动开销」。
+
+## 环境缺口
+
+- 本机 `.venv`：`sys.executable` 是工作树的 `python.exe`；`shutil.which("python3")` 是 Windows 应用商店占位；`python3.14` 没有；`find_spec("jhora")` 是 `None`。
+- Docker 在，没有本地 `jyotisha-api` 镜像，镜像里的解释器路径没有核实。
+- 生产装了 VedAstro、并且开了外网时，不带 skip 的排盘真实耗时本机测不到。
+- 名额被占满后连带拒绝的真实频率，要等日志上了 staging 再看。
+
+## 需拍板
+
+轻重任务要不要分开限流，本单不改。模拟里，让咨询少被拒，校正就会多被拒；给咨询单独开池会让同时在跑的个数超过 2。要改规则需要另立一单。
+
+## 默认值
+
+- `JYOTISH_GATE_CHART_OVERVIEW` 默认关。
+- 合盘 skip 和报告短路没有开关。
+
+## 2026-10-15 打回修复
+
+合盘路由的合同跟上了新写法：仍锁本人和对方各请求一次排盘，并锁住关闭外部概览。对照脚本默认写到系统临时目录，只有带 `--out` 才写指定文件，不再覆盖已入库的记录。业务代码没改。名额配额仍不实现。没有新 bug 编号，补在 BUG-1333 的两处副本上。
+
+| 检查 | 命令 | 结果 |
+| --- | --- | --- |
+| 用户数据合同 | `.venv\Scripts\python.exe -m pytest tests/test_supabase_user_data_contract.py -q` | 8 passed，退出码 0 |
+| 脚本默认路径 | 调用 `resolve_record_path` | 默认不在仓库内，也不是已入库的 json；`--out` 指向传入路径 |
+| 源码文本检索 | `tests` 与 `frontend/tests` 里的合盘路由、`postPython("/api/chart"`、`postPython("/api/varga_full"` | 只有这一条合同在断言旧写法，已改。其它命中是在读路由文件，没有断言旧的两次调用。分盘前缀仍匹配，没改 |
+| 类型检查 | 在 `frontend` 下 `node_modules\.bin\tsc --noEmit -p .` | 退出码 0 |
+| lint | 在 `frontend` 下 `npm run lint` | 退出码 0，0 error，123 条原有 warning |
+
+原值：`postPython("/api/chart", selfPayload)` 与 `postPython("/api/chart", partnerPayload)`。新值：两次调用都展开对应 payload，并带 `skip_vedastro_main_entry_overview: true`。原因：合盘改为同时排两张盘并关掉用不上的外部概览，合同要锁住这两次请求和这个开关，不放宽。
+
+偏离：先合并了 `origin/staging` `1de3cc307be9134a3b8d1d606c738c1468414118`。相对打回 SHA `336286e4`，`frontend/src` 里多出来的是数据访问层那一单，不是这次修复改的。技能目录下的同名脚本没改，它的输出根不是仓库的 `docs/testing`。
diff --git a/docs/tasks/PROGRESS-beta-safety-net-20261014.md b/docs/tasks/PROGRESS-beta-safety-net-20261014.md
index b174979b..b2d37131 100644
--- a/docs/tasks/PROGRESS-beta-safety-net-20261014.md
+++ b/docs/tasks/PROGRESS-beta-safety-net-20261014.md
@@ -3,7 +3,7 @@
 - 分支：`codex/beta-safety-net-20261014`
 - 基线：开工时 `origin/staging` 为 `71f7fac3a13f817f91755ff01b25b8706380ae19`
 - 前置：`09906cdc`（发布通路）和 `4c3485bc`、`3893783e`（#58）都已是这个基线的祖先。`deploy/docker-compose.server.yml` 在加日志段之前没有 #58 留下的未合入冲突。
-- 编号：本文件最大号是 BUG-1353。本单用 BUG-1354（没有每日备份）、BUG-1355（没有探活和磁盘告警）、BUG-1356（日志不轮转、不能按请求编号查）。
+- 编号：本文件最大号是 BUG-1353。本单当时写成 BUG-1354（没有每日备份）、BUG-1355（没有探活和磁盘告警）、BUG-1356（日志不轮转、不能按请求编号查）。合并 staging 后 BUG-1354 已被更早的答题强度记录占用，三条顺延为 BUG-1355、BUG-1356、BUG-1357。
 
 ## 做了什么
 
diff --git a/docs/tasks/README.md b/docs/tasks/README.md
index baf4927e..b31d3ab7 100644
--- a/docs/tasks/README.md
+++ b/docs/tasks/README.md
@@ -338,8 +338,10 @@
 | `TASK-release-path-production-20261014.md` | `PROGRESS-release-path-production-20261014.md` | **发布通路 + 生产首发**：产品 10-14 授权改生产工作流（发布 SHA = 最近门禁成功提交 + 其后纯文档）；63 个迁移清单与演练脚本、首发 runbook；生产拒绝测试验证码。Grok 不 dispatch 生产工作流 | 第 2 次打回后续修 | `codex/release-path-production-20261014`（BUG-1348、BUG-1349） |
 | `TASK-release-path-production-fix-20261014.md` | `PROGRESS-release-path-production-fix-20261014.md` | **#74 第 1 次打回**：发布规则成立；测试写死 Windows 路径（Linux 全红）、主机演练依赖主机 node 且会留下生产数据副本、docs-only 头时回滚被拒；另补 CHECK_RELEASE_GATE 断言与 OTP 守卫规范化 | 已修，第 2 次仍未过 | `codex/release-path-production-20261014` |
 | `TASK-release-path-production-fix2-20261014.md` | `PROGRESS-release-path-production-fix2-20261014.md` | **#74 修复单 2（产品选 A）**：恢复点要带属主与权限，旧格式不得回退；演练按备份里的授权比对，并在真实 PostgreSQL 17 上跑通一次 | 待验收 | `codex/release-path-production-20261014` |
-| `TASK-beta-safety-net-20261014.md` | `PROGRESS-beta-safety-net-20261014.md` | **内测安全网**：每日备份 + 恢复演练、探活告警、本地错误汇总、日志轮转、按 requestId 查日志。发布通路与 #58 已在开工时的 `origin/staging` 上。主机安装和一次演练留给产品 | 待验收 | `codex/beta-safety-net-20261014`（BUG-1354～1356） |
+| `TASK-beta-safety-net-20261014.md` | `PROGRESS-beta-safety-net-20261014.md` | **内测安全网**：每日备份 + 恢复演练、探活告警、本地错误汇总、日志轮转、按 requestId 查日志。发布通路与 #58 已在开工时的 `origin/staging` 上。主机安装和一次演练留给产品 | 待验收 | `codex/beta-safety-net-20261014`（BUG-1355～1357） |
+| `TASK-beta-safety-net-fix-20261015.md` | — | **#75 产品选 A**：授权新增每日备份定时工作流，但每日备份不得改写 `recovery-baseline.env`（恢复点仍只手动创建）；另记 `daily-backup-latest.env` | 待领取 | `codex/beta-safety-net-20261014` |
 | `TASK-beta-capacity-20261014.md` | — | **内测容量**：按用户限流、重计算短排队与对话重试、兑换码当邀请码、staging 10/20 并发压测。**等 #58 合入后开工，压测等 #57 合入** | 待前置 | — |
+| `TASK-beta-capacity-fix-20261015.md` | — | **#76 第 1 次打回**：`a54f75e5` 限流与短排队逻辑通过；源码文本断言未跟改、插入新测试使隐私白名单行号错位 | 待领取 | `codex/beta-capacity-20261014` |
 | `TASK-special-lagnas-native-20261008.md` | `PROGRESS-special-lagnas-native-20261008.md` | **特殊上升点原生计算（线上缺陷）**：sync6 U6 运行时调 PyJHora，生产镜像没有 PyJHora → staging 上 Hora/Ghati/Sree/Indu/Pranapada 全 blocked；原生实现、PyJHora 只作冻结对照、快速门禁止运行时 import jhora；最先做 | 已验收（已部署 staging `795b8845`，BUG-1273） | `codex/special-lagnas-native-20261008`（未上线；部署以 `/api/health` 为准） |
 | `TASK-ephemeris-nutation-20261008.md` | `PROGRESS-ephemeris-nutation-20261008.md` | **恒星黄道少扣章动（BUG-1275）**：行星 = 含章动热带视位置 − 平岁差，全体偏 Δψ（≤17″），大运日期差 2–3 天；改标准算法、PyJHora 冻结对照、scoring-12、v5 重跑 | 已验收（已部署 staging `db0134f9`，BUG-1275） | `codex/ephemeris-nutation-20261008`（未推送，未部署） |
 | `TASK-sync6-gate-red-20261007.md` | — | **sync6 后门禁仍红**：run 3213 快速门已过，「Frontend and database tests」开跑 20 秒后日志中断、15 分钟后判失败，后续步骤 failure 0 秒；疑运行机掉线；先 Re-run，红则本机带 Docker 跑 `CI=1 npm test` 对照 `0ed45a4b` | 已完成（Re-run 通过，BUG-1271，部署 `2df6ee72`） | — |
@@ -483,10 +485,10 @@
 | `TASK-consult-prep-parallel-fix-20261014.md` | `PROGRESS-consult-prep-parallel-20261013.md` | **#62 第 1 次打回**：两条源码文本检查改认新位置，对照脚本挪进测试目录 | 待验收 | `codex/consult-prep-parallel-20261013` |
 | `TASK-frontend-dead-code-20261013.md` | `PROGRESS-frontend-dead-code-20261013.md` | 前端死代码 17 文件与无调用方接口删除，含未登录重计算接口（BUG-1342） | 待验收 | `codex/frontend-dead-code-20261013` |
 | `TASK-frontend-dead-code-fix-20261014.md` | `PROGRESS-frontend-dead-code-20261013.md` | **#63 第 1 次打回**：类名白名单去掉已删组件留下的条目 | 待验收 | `codex/frontend-dead-code-20261013` |
-| `TASK-api-chart-subprocess-gate-20261013.md` | — | 合盘不带 skip 串行排盘、报告无效子进程、名额轻重混用（只模拟）（BUG-1333～1334，A/B） | 待领取（排队：api-disk-cache-retention + report-reference-perf） | `codex/api-chart-subprocess-gate-20261013` |
-| `TASK-api-chart-subprocess-gate-fix-20261015.md` | — | **#64 第 1 次打回**：`336286e4` 内容通过（合盘 1.19→0.015 s、正文逐字节一致）；Python 合同按源码检查合盘路由旧写法未跟改；A/B 脚本默认覆盖入库记录 | 待领取 | `codex/api-chart-subprocess-gate-20261013` |
-| `TASK-answer-effort-ab-20261013.md` | — | 答题步推理强度开关 + A/B（需模型 key） | 待领取（排队：consult-prep-parallel） | `codex/answer-effort-ab-20261013` |
-| `TASK-answer-effort-ab-fix-20261015.md` | — | **#65 第 1 次打回**：`ab7f66be` 开关逻辑通过（默认请求体不变）；研究脚本从 `frontend/tests` 读文件（BUG-1140） | 待领取 | `codex/answer-effort-ab-20261013` |
+| `TASK-api-chart-subprocess-gate-20261013.md` | `PROGRESS-api-chart-subprocess-gate-20261013.md` | 合盘不带 skip 串行排盘、报告无效子进程、名额轻重混用（只模拟）（BUG-1333～1334，A/B） | 第 1 次打回，见修复单 | `codex/api-chart-subprocess-gate-20261013` |
+| `TASK-api-chart-subprocess-gate-fix-20261015.md` | `PROGRESS-api-chart-subprocess-gate-20261013.md` | **#64 第 1 次打回**：`336286e4` 内容通过（合盘 1.19→0.015 s、正文逐字节一致）；Python 合同按源码检查合盘路由旧写法未跟改；A/B 脚本默认覆盖入库记录 | 待验收 | `codex/api-chart-subprocess-gate-20261013` |
+| `TASK-answer-effort-ab-20261013.md` | `PROGRESS-answer-effort-ab-20261013.md` | 答题步推理强度开关 + A/B（需模型 key） | 待验收 | `codex/answer-effort-ab-20261013` |
+| `TASK-answer-effort-ab-fix-20261015.md` | `PROGRESS-answer-effort-ab-fix-20261015.md` | **#65 第 1 次打回**：`ab7f66be` 开关逻辑通过（默认请求体不变）；研究脚本从 `frontend/tests` 读文件（BUG-1140），已挪到 `frontend/tests/answer-effort-probe.mts` | 待验收 | `codex/answer-effort-ab-20261013` |
 | `TASK-consult-route-split-20261013.md` | — | consult POST 拆分第一阶段：开流前加载 + 计费结算 | 待领取（排队：consult-prep-parallel、answer-effort-ab） | `codex/consult-route-split-20261013` |
 | `TASK-pl9-render-split-20261013.md` | — | render_pl9_markdown 第一阶段：搬出无依赖内部函数 | 待领取（排队：api-chart-subprocess-gate） | `codex/pl9-render-split-20261013` |
 | `TASK-python-hygiene-20261013.md` | — | 免费层排队开关报反、静默吞异常、功能吉凶兜底合一、测试专用脚本归位（BUG-1340） | 待领取（排队：pl9-render-split） | `codex/python-hygiene-20261013` |
diff --git a/docs/tasks/TASK-beta-capacity-fix-20261015.md b/docs/tasks/TASK-beta-capacity-fix-20261015.md
new file mode 100644
index 00000000..07c298db
--- /dev/null
+++ b/docs/tasks/TASK-beta-capacity-fix-20261015.md
@@ -0,0 +1,37 @@
+# TASK · 内测容量修复单（issue #76，第 1 次打回）
+
+- 日期：2026-10-15
+- 原任务书：`docs/tasks/TASK-beta-capacity-20261014.md`
+- 被验 SHA：`a54f75e5e33775d5dc91f3058f5505a56b498bf2`（`codex/beta-capacity-20261014`）
+- 执行方：Grok，同一分支续做（先 merge 最新 `origin/staging`，不改写已推历史）
+
+## 1. 验收结论：逻辑通过，全量多出 2 条失败
+
+| 检查 | 结果 | 证据 |
+| --- | --- | --- |
+| 按用户限流 | 通过 | 问、校正每轮、报告每小时各有分钟 / 小时两档；任一档满就整体拒，拒绝不记账；都在扣点之前 |
+| 重计算短排队 | 通过 | 并发仍是 2（BUG-717）；最多 4 个等待、每个最多 8 秒，等不到仍 429（BUG-720 有上限） |
+| 前端排队重试 | 通过 | 最多 2 次、总等待 ≤ 15 秒；最终失败走原取消路径 |
+| tsc / lint | 通过 | — |
+| **前端全量** | **未通过：新增 1 条** | `frontend/tests/consultation-voice-contract.test.ts`「consult user turn no longer dumps tool JSON that the model would parrot」：`/await runConsultationWorkflow\(toolInput/` 不再匹配——`route.ts` 改成了 `await retryOnConsultationRateLimit({ run: () => runConsultationWorkflow(toolInput, …) })` |
+| **Python 全量** | **未通过：新增 1 条** | `tests/test_repo_privacy_markers.py::test_tracked_repository_has_no_privacy_markers`：`HOME_PREFIX frontend/tests/personal-report-api.test.ts lines=845,873,903` |
+
+第 2 条不是真的隐私泄漏：这三行是已有的、经审过的虚构路径样本，`tests/test_repo_privacy_markers.py` 的 `SAFE_LINES` 按「文件 + 行号 + 整行」锁定它们。本单在它们上面插入了一条 13 行的新测试，行号整体下移，白名单就对不上了。
+
+## 2. 修法要求
+
+1. `consultation-voice-contract.test.ts`：按 §7.3 写三栏改断言，**意图不放宽**——仍要锁住「工作流调用的入参是 `toolInput`」与该测试原本关心的那段（用户轮不再塞工具 JSON）；新正则匹配 `runConsultationWorkflow(toolInput` 即可，不得去掉对 `toolInput` 的锁定。
+2. 隐私白名单二选一（推荐 a）：
+   a. 把新加的「hourly rate limit returns 429」测试挪到文件末尾，让那三行回到原行号，`SAFE_LINES` 不动；
+   b. 更新 `SAFE_LINES` 里这三行的行号，按三栏写原值 / 新值 / 原因。
+3. 再 grep 一遍 `route.ts` / `rectification/agent/route.ts` / `reports/route.ts` 被改写语句在 `frontend/tests`、`tests` 里的源码文本断言（§7.8）。
+4. 交回前本机跑：`npx tsx --test tests/consultation-voice-contract.test.ts tests/runtime-sources-no-test-imports.test.ts` 与 `python -m pytest tests/test_repo_privacy_markers.py -q`。
+5. 业务代码不改。
+
+## 3. 环境缺口（不阻断）
+
+`frontend/scripts/research/beta-load-smoke.mts` 并发压测未跑（无测试验证码、无预发主机）。合入部署后由 Claude 在验收清单里列给产品负责人，或另开一单在 staging 跑。
+
+## 4. 让步顺序
+
+无。第 1 次打回。
diff --git a/docs/tasks/TASK-beta-safety-net-fix-20261015.md b/docs/tasks/TASK-beta-safety-net-fix-20261015.md
new file mode 100644
index 00000000..e6d862aa
--- /dev/null
+++ b/docs/tasks/TASK-beta-safety-net-fix-20261015.md
@@ -0,0 +1,31 @@
+# TASK · 内测安全网修复单（issue #75，产品选 A）
+
+- 日期：2026-10-15
+- 原任务书：`docs/tasks/TASK-beta-safety-net-20261014.md`
+- 被验 SHA：`53d8df257ddef8918034acef36df0b4b7daade2c`（`codex/beta-safety-net-20261014`）
+- 执行方：Grok，同一分支续做（先 merge 最新 `origin/staging`，不改写已推历史）
+- **决策记录**：产品负责人 2026-10-15 在 issue #75 选 **A**——**授权新增定时工作流 `.gitea/workflows/backup-production-daily.yml`**（每天北京时间 04:00 自动连生产做加密备份），但**每日备份不得改动「当前恢复点」记录**。恢复点仍只由产品负责人手动点「Create production recovery」创建。仍不改已有四个生产工作流的语义；仍不点任何生产按钮。
+
+## 1. 问题
+
+`deploy/run-production-recovery.sh` 无论由谁调用，结尾都会无条件写 `$DEPLOY_PATH/.state/recovery-baseline.env`（`RECOVERY_REFERENCE`、`RECOVERY_SHA256`、`RECOVERY_FORMAT=2` 等）。发布、迁移、回退都读这份文件当「当前恢复点」。每日备份复用这个脚本，于是每天 04:00 都会把它换成当天的自动备份——产品负责人上线前手动建的恢复点，隔一夜就被替换。
+
+## 2. 修法要求
+
+1. **每日备份不写 `recovery-baseline.env`**：`RECOVERY_BASENAME_PREFIX=production-daily` 时（或新增一个明确的模式参数，二选一，写进 PROGRESS），脚本仍完整执行「导出 → 加密 → 还原核对（含 #74 的表权限、列权限、属主三项比对）→ 写 `*-restore-verify.json`」，但**跳过写 `recovery-baseline.env`**，并在输出里打印 `baseline_updated=false`。手动恢复点路径行为不变，打印 `baseline_updated=true`。
+2. 每日备份的核对结果单独记录：在 `$DEPLOY_PATH/.state/` 下另写一个 `daily-backup-latest.env`（只含文件名、sha256、校验时间、格式版本，不含数据、不含密钥），供告警 / 运维查看；发布、迁移、回退工作流**不得**读取它。
+3. 合同测试（`tests/test_beta_safety_net.py` 或 `frontend/tests/staging-backend-workflows.test.ts`）：
+   - 每日模式下 `recovery-baseline.env` 不被创建或修改（用临时目录 + 假 docker 跑脚本，断言文件 mtime / 内容不变）；
+   - 手动模式下仍会写，且含 `RECOVERY_FORMAT=2`；
+   - 三个读取恢复点的生产工作流 / 脚本不引用 `daily-backup-latest.env`。
+4. runbook 与 `deploy/README.md`：写清「每日备份只做存档与核对，不是发布用的恢复点；上线前仍要手动点 Create production recovery」。
+5. 清理脚本只删 `production-daily-*` 前缀的文件（原任务书已要求）——再确认不会删 `production-pre-migration-*` 手动恢复点。
+
+## 3. 验收
+
+- Claude 会对本单做完整验收（上一轮在发现定时工作流后就停下等拍板，其余部分尚未逐行验）：逐行看新工作流与各脚本、跑 `tests/test_beta_safety_net.py`、Python 全量、前端全量、`docker compose config`。
+- 交回前本机跑：`python -m pytest tests/test_beta_safety_net.py tests/test_repo_privacy_markers.py -q`、`npx tsx --test tests/runtime-sources-no-test-imports.test.ts tests/staging-backend-workflows.test.ts`。
+
+## 4. 让步顺序
+
+无。
diff --git a/docs/testing/answer-effort-ab-20261013.md b/docs/testing/answer-effort-ab-20261013.md
new file mode 100644
index 00000000..a476c414
--- /dev/null
+++ b/docs/testing/answer-effort-ab-20261013.md
@@ -0,0 +1,43 @@
+# 答题步推理强度（2026-10-15）
+
+没有模型钥匙。下面只是本机拦截到的请求体，不是供应商的接受度，也不是耗时和质量。A/B 未跑。
+
+命令：在 `frontend` 目录执行 `node node_modules/tsx/dist/cli.mjs tests/answer-effort-probe.mts --dry-run`。假请求不离开这台机器。模型对象与产品相同：`providerId: deepseek`，`modelId: deepseek-flash`，带 `url` 和假的 `apiKey`。2026-10-15 起脚本在 `frontend/tests/`，不再放在 `frontend/scripts/research/`。
+
+## 请求体里实际有什么
+
+| 形状 | thinking | reasoning_effort | max_tokens |
+| --- | --- | --- | --- |
+| 今天的答题设置，第 0 步和第 1 步 | `{ type: enabled }` | 无 | 24576 |
+| 旧的低强度设置（没有产品调用方） | `{ type: enabled, reasoningEffort: low }` | 无 | 10240 |
+| 新开关 `low` / `medium` / `high` / `max`（只在答题步） | `{ type: enabled }` | 与开关相同 | 24576 |
+| 续写 | `{ type: disabled }` | 无 | 16384 |
+
+旧的低强度写在 `thinking` 里面，请求体没有 `reasoning_effort`。这证实了任务书 §2.3。记为 BUG-1354，状态 investigating。原先写成 BUG-1351，该号已被数据层占用。线上不用这处设置。
+
+新开关写在供应商选项的最外层。兼容通道把它写成 `reasoning_effort`。开关不设时，两步请求体与改前逐字节一致，样本在 `frontend/tests/fixtures/answer-effort-switch-20261013.json`。
+
+## 供应商接受度
+
+未跑。没有 `DS_KEY` / `BACKTEST_MODEL_KEY`。`low`、`medium`、`high`、`max` 哪几档能用，要等验收时用临时钥匙跑探针的后半段。
+
+## 固定 10 题
+
+脚本已能按 `--cases` 跑下面这 10 题，并能按 `--efforts unset,low,medium` 把同一题的几档挨着跑。多档同时跑时并发降为 1，因为强度读的是进程里的环境变量，并行会串档。本机没有钥匙，这 10 题没有跑。
+
+| # | 盘 | 领域 |
+| --- | --- | --- |
+| 1 | steve_jobs | career |
+| 2 | barack_obama | parents |
+| 3 | elizabeth_taylor | marriage |
+| 4 | marilyn_monroe | parents |
+| 5 | judy_garland | health |
+| 6 | edith_piaf | marriage |
+| 7 | frida_kahlo | health |
+| 8 | george_w_bush | career |
+| 9 | zinedine_zidane | career |
+| 10 | marilyn_monroe | marriage |
+
+## 耗时与质量
+
+没有数。空答、截断、rubric、人工四项都没做。默认值维持不设。
diff --git a/docs/testing/api-chart-subprocess-gate-ab-20261013.json b/docs/testing/api-chart-subprocess-gate-ab-20261013.json
new file mode 100644
index 00000000..44c6b5be
--- /dev/null
+++ b/docs/testing/api-chart-subprocess-gate-ab-20261013.json
@@ -0,0 +1,53 @@
+{
+  "charts": {
+    "serial_seconds_median": 8.0596,
+    "concurrent_skip_seconds_median": 0.0166,
+    "serial_subprocesses_median": 2,
+    "concurrent_skip_subprocesses_median": 0,
+    "summary_bytes_match": true
+  },
+  "report_forced_subprocess": {
+    "seconds_median": 16.5322,
+    "subprocesses_median": 7
+  },
+  "report_short_circuit": {
+    "seconds_median": 15.3235,
+    "subprocesses_median": 1
+  },
+  "report_bytes_match": true,
+  "schemes": [
+    {
+      "scheme": "O1",
+      "consult_429": "0.40",
+      "report_429": "0.00",
+      "score_429": "0.17",
+      "peak": 2,
+      "touches_bug_717": false
+    },
+    {
+      "scheme": "O2",
+      "consult_429": "0.00",
+      "report_429": "0.00",
+      "score_429": "0.17",
+      "peak": 3,
+      "touches_bug_717": true
+    },
+    {
+      "scheme": "O3",
+      "consult_429": "0.30",
+      "report_429": "0.00",
+      "score_429": "0.33",
+      "peak": 2,
+      "touches_bug_717": false
+    },
+    {
+      "scheme": "O4",
+      "consult_429": "0.00",
+      "report_429": "0.00",
+      "score_429": "0.67",
+      "peak": 2,
+      "touches_bug_717": false
+    }
+  ],
+  "jhora_installed": false
+}
diff --git a/docs/testing/api-chart-subprocess-gate-ab-20261013.md b/docs/testing/api-chart-subprocess-gate-ab-20261013.md
new file mode 100644
index 00000000..55a88cb8
--- /dev/null
+++ b/docs/testing/api-chart-subprocess-gate-ab-20261013.md
@@ -0,0 +1,40 @@
+# 合盘排盘与报告子进程 A/B（2026-10-15）
+
+Windows 本机，工作树 `.venv` 的 `python.exe`。`importlib.util.find_spec("jhora")` 是 `None`。`python3` 指向 Windows 应用商店占位程序，`python3.14` 不存在。Docker 守护进程在，本机没有 `jyotisha-api` 镜像，所以没有进镜像再探一次。本机没有 VedAstro SDK，外网关闭（`VEDASTRO_ENABLE_NETWORK=0`）。没有「真 SDK」一组。
+
+脚本：`scripts/research/api_chart_subprocess_ab.py`。每组 3 次，取中位数。盘是任务书里的虚构盘（1990-01-01 12:00 与 1992-01-02 12:00，新德里）。星盘缓存 TTL 设为 0。易变字段在比对前丢掉：`generated_at`、`created_at`、`timestamp`、`elapsed_seconds`、`elapsed_ms`、`duration_ms`、`request_id`。原始数字在同目录 `api-chart-subprocess-gate-ab-20261013.json`。
+
+对照不是另一次 `git show` 出 staging 再导入。星盘对照是本分支的 `_compute_chart_sync`：不带 skip、两张盘串行。报告对照是本分支把「参照引擎已安装」强制打开，从而仍会起子进程；静态校验缓存和解释器查找缓存两边都开着。所以报告耗时差只含那 6 个注定失败的回放进程，不含静态校验少跑的那一次。静态校验「连调两次只跑一次」由单测覆盖。
+
+## 合盘两张盘
+
+| | 中位耗时 | 中位 `subprocess.run` 次数 | 盘面摘要 |
+| --- | --- | --- | --- |
+| 对照：不带 skip，串行 | 8.0596 秒 | 2 | — |
+| 本单：带 skip，并发 | 0.0166 秒 | 0 | 与对照逐字节一致 |
+
+0.0166 / 8.0596 ≈ 0.2%，低于「不超过对照 10%」的线。子进程从 2 降到 0。任务书写的「4 → 0」是按每张盘 2 个子进程、两张合计 4 个来估的。这台机器上两张盘合计只数到 2 次 `subprocess.run`。不带 skip 的 8 秒里，大部分是没装 SDK 时子进程空等，不是盘面计算本身。
+
+## 个人长报告（完整数据版，中英，带事实表）
+
+| | 中位耗时 | 中位 `subprocess.run` 次数 | 中英正文与事实表 |
+| --- | --- | --- | --- |
+| 对照：强制仍起回放子进程 | 16.5322 秒 | 7 | — |
+| 本单：没装参照引擎就不起 | 15.3235 秒 | 1 | 去掉易变字段后逐字节一致 |
+
+少 6 次，对应年度回放 3 次和 Tajika 3 次。剩下的 1 次两边都有，短路没有把它去掉。子进程数 1，低于「模拟生产不超过 3」的线。耗时 15.32 秒，不比对照的 16.53 秒差。本机没装 PyJHora，没有再跑「装了参照引擎」的一组；那种环境仍必须走子进程，单测用桩覆盖。
+
+生产上的收益仍按任务书：大约每份报告 0.5 到 0.9 秒，不是装了参照引擎的机器上看到的约 5 秒。
+
+## 名额方案模拟（没改规则）
+
+固定一分钟：2 份报告（每次 13 秒）、20 次咨询（每次 0.2 秒）、6 次校正打分（每次 8 秒）。随机种子 64。并发上限按现状 2。O2、O3、O4 只在脚本里模拟。
+
+| 方案 | 咨询被拒 | 报告被拒 | 校正被拒 | 同时在跑的最高个数 | 是否碰到「不得加池、不得提高并发」 |
+| --- | --- | --- | --- | --- | --- |
+| O1 现状 | 40% | 0% | 17% | 2 | 否 |
+| O2 咨询单独小池 | 0% | 0% | 17% | 3 | 是，本单不实现 |
+| O3 咨询最多再等 1.5 秒 | 30% | 0% | 33% | 2 | 否，本单不实现 |
+| O4 重任务最多占 1 个名额 | 0% | 0% | 67% | 2 | 否，本单不实现 |
+
+这是一分钟的草图，不是 staging 上的真实频率。咨询少被拒的时候，校正会多被拒（O3、O4）。O2 咨询不再被拒，但同时在跑的个数到了 3，越过了现在的上限。
diff --git a/frontend/scripts/research/consult-biography-backtest.mts b/frontend/scripts/research/consult-biography-backtest.mts
index 2083df7f..56ed30bd 100644
--- a/frontend/scripts/research/consult-biography-backtest.mts
+++ b/frontend/scripts/research/consult-biography-backtest.mts
@@ -27,6 +27,7 @@
  * Usage (from frontend/, Node 22):
  *   DS_KEY=... ./node_modules/.bin/tsx scripts/research/consult-biography-backtest.mts \
  *     [--out <dir>] [--repeats 2] [--figures a,b] [--domains parents,marriage,health,career] \
+ *     [--cases steve_jobs:career,barack_obama:parents] [--efforts unset,low,medium] \
  *     [--model deepseek-flash] [--base-url https://api.deepseek.com] [--concurrency 4] [--dump-cards]
  *     [--correction barack_obama,marilyn_monroe] [--correction-text 我爸其实没怎么管过我]
  *     [--correction-domain career] [--correction-texts "zinedine_zidane=我是足球运动员|barack_obama=我是政治家"]
@@ -142,6 +143,8 @@ function parseArgs(argv: readonly string[]) {
       const at = pair.indexOf("=");
       return [pair.slice(0, at), pair.slice(at + 1)];
     })) as Record<string, string>,
+    cases: value("cases")?.split(",").map((item) => item.trim()).filter(Boolean) ?? null,
+    efforts: value("efforts")?.split(",").map((item) => item.trim()).filter(Boolean) ?? null,
   };
 }
 
@@ -270,7 +273,15 @@ function model(args: ReturnType<typeof parseArgs>, apiKey: string): ResolvedLang
 
 /** A follow-up turn in the same session: the earlier question and answer are the history. */
 type FollowUp = { priorQuestion: string; priorAnswer: string };
-type Job = { figure: GoldenFigure; domain: string; question: string; run: number; followUp?: FollowUp };
+type Job = {
+  figure: GoldenFigure;
+  domain: string;
+  question: string;
+  run: number;
+  followUp?: FollowUp;
+  /** When set, this job overrides CONSULT_ANSWER_REASONING_EFFORT for its own stream. */
+  effort?: string;
+};
 
 type Usage = { inputTokens?: number; outputTokens?: number; reasoningTokens?: number; cachedInputTokens?: number };
 
@@ -316,6 +327,22 @@ function userTurn(question: string, followUp?: FollowUp) {
 }
 
 async function runJob(job: Job, resolved: ResolvedLanguageModel) {
+  const previousEffort = process.env.CONSULT_ANSWER_REASONING_EFFORT;
+  if (job.effort !== undefined) {
+    if (job.effort === "unset" || job.effort === "") delete process.env.CONSULT_ANSWER_REASONING_EFFORT;
+    else process.env.CONSULT_ANSWER_REASONING_EFFORT = job.effort;
+  }
+  try {
+    return await runJobOnce(job, resolved);
+  } finally {
+    if (job.effort !== undefined) {
+      if (previousEffort === undefined) delete process.env.CONSULT_ANSWER_REASONING_EFFORT;
+      else process.env.CONSULT_ANSWER_REASONING_EFFORT = previousEffort;
+    }
+  }
+}
+
+async function runJobOnce(job: Job, resolved: ResolvedLanguageModel) {
   const state = createConsultationRuntimeState();
   const executed: string[] = [];
   const ctx = agentContext(job, state, executed);
@@ -333,7 +360,14 @@ async function runJob(job: Job, resolved: ResolvedLanguageModel) {
     maxSteps: AGENT_MAX_STEPS,
     ...consultationGenerationSettings(resolved.model),
   };
-  const natalStreamOptions = { ...streamOptions, prepareStep: consultationNatalPrepareStep };
+  const natalStreamOptions = {
+    ...streamOptions,
+    prepareStep: (input: { stepNumber: number }) => consultationNatalPrepareStep({
+      stepNumber: input.stepNumber,
+      answerReady: state.consultationToolCompleted,
+      model: resolved.model,
+    }),
+  };
   const usages: Promise<unknown>[] = [];
   const startedAt = Date.now();
   let output: string | null = null;
@@ -417,6 +451,17 @@ async function runJob(job: Job, resolved: ResolvedLanguageModel) {
     methodologySections: state.methodologySectionCount,
     lookups: state.evidenceLookupCallCount,
     finish: state.composeFinishReason ?? state.modelFinishReason ?? null,
+    effort: process.env.CONSULT_ANSWER_REASONING_EFFORT?.trim() || null,
+    answerReasoningMs: state.answerReasoningMs ?? null,
+    steps: [0, 1].map((index) => {
+      const step = state.modelStepTimings?.find((item) => item.index === index);
+      return {
+        index,
+        seconds: step?.durationMs == null ? null : step.durationMs / 1000,
+        reasoningTokens: step?.reasoningTokens ?? null,
+        outputTokens: step?.outputTokens ?? null,
+      };
+    }),
   };
 }
 
@@ -477,7 +522,18 @@ console.info = (...items: unknown[]) => {
   productInfo(...items);
 };
 
-type Row = { figure: string; domain: string; run: number; chars: number; seconds: number; checks: ReturnType<typeof deterministicChecks> };
+type Row = {
+  figure: string;
+  domain: string;
+  run: number;
+  chars: number;
+  seconds: number;
+  checks: ReturnType<typeof deterministicChecks>;
+  empty?: boolean;
+  truncated?: boolean;
+  finish?: string | null;
+  steps?: { seconds: number | null; reasoningTokens: number | null; outputTokens: number | null }[];
+};
 
 function writeSummary(out: string, summary: Json & { rows: Row[] }, modelId: string) {
   const rows = summary.rows;
@@ -488,13 +544,13 @@ function writeSummary(out: string, summary: Json & { rows: Row[] }, modelId: str
   writeFileSync(join(out, "summary.md"), [
     `# 名人生平回测 · ${modelId}`,
     "",
-    `开始 ${summary.started_at}，${rows.length} 份，墙钟 ${Number(summary.wall_seconds).toFixed(0)} s，失败 ${summary.failures} 份。`,
+    `开始 ${summary.started_at}，${rows.length} 份，墙钟 ${Number(summary.wall_seconds).toFixed(0)} s，失败 ${summary.failures} 份。推理强度：${String(summary.reasoning_effort ?? "不设")}。`,
     `用量合计：${JSON.stringify(summary.usage)}`,
     `确定性检查：禁句 ${summary.banned_hits}，性别代词（婚恋/子女）${summary.pronoun_hits}，must_not 原文 ${summary.must_not_literal_hits}。判断类评分需逐份阅读。`,
     "",
-    "| 名人 | 领域 | run | 字数 | 秒 | 禁句 | 代词 | must_not |",
-    "| --- | --- | --- | --- | --- | --- | --- | --- |",
-    ...rows.map((row) => `| ${row.figure} | ${row.domain} | ${row.run} | ${row.chars} | ${row.seconds.toFixed(0)} | ${row.checks.banned.length} | ${row.checks.pronouns.length} | ${row.checks.literalHits.length} |`),
+    "| 名人 | 领域 | run | 字数 | 秒 | 空答 | 截断 | 第0步秒 | 第1步秒 | 第1步推理token | 结束 |",
+    "| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |",
+    ...rows.map((row) => `| ${row.figure} | ${row.domain} | ${row.run} | ${row.chars} | ${row.seconds.toFixed(0)} | ${row.empty ? "是" : ""} | ${row.truncated ? "是" : ""} | ${row.steps?.[0]?.seconds ?? ""} | ${row.steps?.[1]?.seconds ?? ""} | ${row.steps?.[1]?.reasoningTokens ?? ""} | ${row.finish ?? ""} |`),
     "",
     "## must_not 原文命中",
     "",
@@ -529,11 +585,28 @@ async function main() {
   }
   const questions = new Map(golden.routes.map((route) => [route.domain, route.question]));
   const figures = golden.figures.filter((figure) => !args.figures || args.figures.includes(figure.id));
-  const jobs: Job[] = figures.flatMap((figure) => args.domains.flatMap((domain) => {
-    const question = questions.get(domain);
-    if (!question) throw new Error(`no golden question for domain ${domain}`);
-    return Array.from({ length: args.repeats }, (_, index) => ({ figure, domain, question, run: index + 1 }));
-  }));
+  const pairs = args.cases
+    ? args.cases.map((spec) => {
+      const splitAt = spec.indexOf(":");
+      const figureId = spec.slice(0, splitAt);
+      const domain = spec.slice(splitAt + 1);
+      const figure = golden.figures.find((item) => item.id === figureId);
+      const question = questions.get(domain);
+      if (!figure || !question) throw new Error(`unknown case ${spec}`);
+      return { figure, domain, question };
+    })
+    : figures.flatMap((figure) => args.domains.flatMap((domain) => {
+      const question = questions.get(domain);
+      if (!question) throw new Error(`no golden question for domain ${domain}`);
+      return [{ figure, domain, question }];
+    }));
+  const efforts = args.efforts;
+  const jobs: Job[] = pairs.flatMap((pair) => Array.from({ length: args.repeats }, (_, index) => {
+    const run = index + 1;
+    if (!efforts) return [{ ...pair, run }];
+    return efforts.map((effort) => ({ ...pair, run, effort }));
+  }).flat());
+  const concurrency = efforts && efforts.length > 1 ? 1 : args.concurrency;
   mkdirSync(args.out, { recursive: true });
   if (args.dumpCards) {
     await dumpCards(jobs, args.out);
@@ -543,12 +616,12 @@ async function main() {
   if (!apiKey) throw new Error("Set DS_KEY (or BACKTEST_MODEL_KEY); without a key the backtest is an environment gap.");
   const resolved = model(args, apiKey);
   const started = Date.now();
-  const rows = await pool(jobs, args.concurrency, async (job) => {
+  const rows = await pool(jobs, concurrency, async (job) => {
     let result: Awaited<ReturnType<typeof runJob>>;
     try {
       result = await runJob(job, resolved);
     } catch (error) {
-      result = { output: "", failure: error instanceof Error ? error.message : String(error), executed: [], seconds: 0, usage: {}, cardChars: null, visibleChars: null, methodologySections: 0, lookups: 0, finish: null };
+      result = { output: "", failure: error instanceof Error ? error.message : String(error), executed: [], seconds: 0, usage: {}, cardChars: null, visibleChars: null, methodologySections: 0, lookups: 0, finish: null, effort: job.effort ?? null, answerReasoningMs: null, steps: [{ index: 0, seconds: null, reasoningTokens: null, outputTokens: null }, { index: 1, seconds: null, reasoningTokens: null, outputTokens: null }] };
     }
     const mustNot = rubric.figures.find((item) => item.id === job.figure.id)?.domains[job.domain]?.must_not ?? [];
     const checks = deterministicChecks(result.output, job.domain, mustNot);
@@ -573,7 +646,18 @@ async function main() {
       "",
     ].join("\n"));
     console.log(`${job.figure.id} ${job.domain} run${job.run}: ${result.output.length} chars, ${result.seconds.toFixed(0)} s${result.failure ? `, FAILED ${result.failure}` : ""}`);
-    return { figure: job.figure.id, domain: job.domain, run: job.run, chars: result.output.length, ...result, output: undefined, checks };
+    return {
+      figure: job.figure.id,
+      domain: job.domain,
+      run: job.run,
+      chars: result.output.length,
+      empty: result.output.trim() === "",
+      truncated: result.finish === "length" || (result.failure ?? "").includes("answer_truncated"),
+      outputTokens: result.usage.outputTokens ?? null,
+      ...result,
+      output: undefined,
+      checks,
+    };
   });
   const corrections = await pool(
     jobs.filter((job) => args.correctionDomains.includes(job.domain)
@@ -589,7 +673,7 @@ async function main() {
       try {
         result = await runJob(job, resolved);
       } catch (error) {
-        result = { output: "", failure: error instanceof Error ? error.message : String(error), executed: [], seconds: 0, usage: {}, cardChars: null, visibleChars: null, methodologySections: 0, lookups: 0, finish: null };
+        result = { output: "", failure: error instanceof Error ? error.message : String(error), executed: [], seconds: 0, usage: {}, cardChars: null, visibleChars: null, methodologySections: 0, lookups: 0, finish: null, effort: job.effort ?? null, answerReasoningMs: null, steps: [{ index: 0, seconds: null, reasoningTokens: null, outputTokens: null }, { index: 1, seconds: null, reasoningTokens: null, outputTokens: null }] };
       }
       const file = join(args.out, job.figure.id, `${job.domain}-correction-run${job.run}.md`);
       writeFileSync(file, [
@@ -619,6 +703,9 @@ async function main() {
     jobs: rows.length,
     failures: rows.filter((row) => row.failure || row.chars === 0).length,
     usage: total,
+    reasoning_effort: process.env.CONSULT_ANSWER_REASONING_EFFORT?.trim() || null,
+    efforts: args.efforts,
+    concurrency,
     gaps: GAPS,
     rows,
     corrections,
diff --git a/frontend/src/app/api/consult/route.ts b/frontend/src/app/api/consult/route.ts
index e9e6a056..db0478b4 100644
--- a/frontend/src/app/api/consult/route.ts
+++ b/frontend/src/app/api/consult/route.ts
@@ -75,6 +75,8 @@ import {
   createConsultationRunClock,
   consultationContinueGenerationSettings,
   consultationGenerationSettings,
+  consultationGeneralPrepareStep,
+  consultAnswerEffortProviderOptions,
   consultationNatalPrepareStep,
   consultationWindowPrepareStep,
   createConsultationAgentContext,
@@ -1288,6 +1290,8 @@ export async function POST(request: Request) {
       prepareStep: (input: { stepNumber: number }) => consultationNatalPrepareStep({
         stepNumber: input.stepNumber,
         rectificationOffer: true,
+        answerReady: state.consultationToolCompleted,
+        model: selectedModel.model,
       }),
     };
     const windowStreamOptions = {
@@ -1295,6 +1299,15 @@ export async function POST(request: Request) {
       prepareStep: (input: { stepNumber: number }) => consultationWindowPrepareStep({
         stepNumber: input.stepNumber,
         rectificationOffer: true,
+        answerReady: state.consultationToolCompleted,
+        model: selectedModel.model,
+      }),
+    };
+    const generalStreamOptions = {
+      ...streamOptions,
+      prepareStep: (input: { stepNumber: number }) => consultationGeneralPrepareStep({
+        stepNumber: input.stepNumber,
+        model: selectedModel.model,
       }),
     };
     const answerPhaseSignal = runClock.answerSignal;
@@ -1305,10 +1318,10 @@ export async function POST(request: Request) {
       agent: {
         stream: (
           messages: typeof baseMessages,
-          options: typeof streamOptions | typeof natalStreamOptions | typeof windowStreamOptions,
+          options: typeof streamOptions | typeof natalStreamOptions | typeof windowStreamOptions | typeof generalStreamOptions,
         ) => Promise<{ fullStream: AsyncIterable<unknown> | ReadableStream<unknown>; totalUsage: Promise<Usage> }>;
       },
-      options: typeof streamOptions | typeof natalStreamOptions | typeof windowStreamOptions = streamOptions,
+      options: typeof streamOptions | typeof natalStreamOptions | typeof windowStreamOptions | typeof generalStreamOptions = streamOptions,
     ) {
       try {
         const result = await agent.stream(baseMessages, options);
@@ -1369,14 +1382,14 @@ export async function POST(request: Request) {
     if (usesPublicDailyGeneralAgent(consultationMode, generalDailyContext)) {
       state.workflowReceipt = workflowReceipt;
       const agent = getGeneralJyotishAgent(selectedModel);
-      const result = await streamWithOverflowRetry(agent);
+      const result = await streamWithOverflowRetry(agent, generalStreamOptions);
       // This mode has no calculation to require and no chart method to bind,
       // so there is no contract for a retry to repair.
       const retryForAnswer = async (retryHint?: string) => {
         const retried = await agent.stream([
           ...baseMessages,
           { role: "user" as const, content: `上一轮没有输出任何回答文本。请直接给出这个问题的回答，不要只说明过程。${retryHint ? `\n${retryHint}` : ""}` },
-        ], { ...streamOptions, abortSignal: answerPhaseSignal() });
+        ], { ...generalStreamOptions, abortSignal: answerPhaseSignal() });
         usages.push(retried.totalUsage);
         return retried.fullStream;
       };
@@ -1472,7 +1485,11 @@ export async function POST(request: Request) {
             role: "user" as const,
             content: `服务器窗口计算已经完成，但上一轮没有输出任何回答文本。请重新取回本次计算结果，然后直接给出回答；不要只描述过程或工具调用。${retryHint ? `\n${retryHint}` : ""}`,
           },
-        ], { ...streamOptions, abortSignal: answerPhaseSignal() });
+        ], {
+          ...streamOptions,
+          abortSignal: answerPhaseSignal(),
+          ...consultAnswerEffortProviderOptions(selectedModel.model),
+        });
         usages.push(retried.totalUsage);
         return retried.fullStream;
       };
diff --git a/frontend/src/app/api/synastry/route.ts b/frontend/src/app/api/synastry/route.ts
index af742338..fa657cb4 100644
--- a/frontend/src/app/api/synastry/route.ts
+++ b/frontend/src/app/api/synastry/route.ts
@@ -183,8 +183,10 @@ export async function POST(request: Request) {
       synastryBirthPayload(globalBirthProfileFromAccountRow(selfResolved.row)),
       synastryBirthPayload(globalBirthProfileFromAccountRow(partnerResolved.row)),
     ]);
-    const selfChart = await postPython("/api/chart", selfPayload);
-    const partnerChart = await postPython("/api/chart", partnerPayload);
+    const [selfChart, partnerChart] = await Promise.all([
+      postPython("/api/chart", { ...selfPayload, skip_vedastro_main_entry_overview: true }),
+      postPython("/api/chart", { ...partnerPayload, skip_vedastro_main_entry_overview: true }),
+    ]);
     const relationshipType = parsed.data.relationshipType === "business" || parsed.data.relationshipType === "family" || parsed.data.relationshipType === "general"
       ? parsed.data.relationshipType
       : "romance";
@@ -210,18 +212,20 @@ export async function POST(request: Request) {
         message: "该关系类型尚无可验证的专用合盘计算合同；已保留问题草稿。",
       });
     }
-    const selfD9 = await postPython("/api/varga_full", {
-      ...selfPayload,
-      planets: selfChart.planets,
-      ascendant: selfChart.ascendant,
-      divisions: ["D9"],
-    });
-    const partnerD9 = await postPython("/api/varga_full", {
-      ...partnerPayload,
-      planets: partnerChart.planets,
-      ascendant: partnerChart.ascendant,
-      divisions: ["D9"],
-    });
+    const [selfD9, partnerD9] = await Promise.all([
+      postPython("/api/varga_full", {
+        ...selfPayload,
+        planets: selfChart.planets,
+        ascendant: selfChart.ascendant,
+        divisions: ["D9"],
+      }),
+      postPython("/api/varga_full", {
+        ...partnerPayload,
+        planets: partnerChart.planets,
+        ascendant: partnerChart.ascendant,
+        divisions: ["D9"],
+      }),
+    ]);
     const synastry = await postPython("/api/synastry", {
       male_moon: moonLongitude(selfChart),
       female_moon: moonLongitude(partnerChart),
diff --git a/frontend/src/lib/agent-generation-settings.ts b/frontend/src/lib/agent-generation-settings.ts
index 25488402..7d3de3f6 100644
--- a/frontend/src/lib/agent-generation-settings.ts
+++ b/frontend/src/lib/agent-generation-settings.ts
@@ -16,6 +16,8 @@ export const AGENT_MAX_OUTPUT_TOKENS = AGENT_ANSWER_OUTPUT_TOKENS;
 
 export type ThinkingMode = "enabled" | "disabled";
 export type ReasoningEffort = "low" | "medium" | "high";
+/** Top-level OpenAI-compatible `reasoning_effort`, including DeepSeek's `max`. */
+export type ProviderReasoningEffort = "low" | "medium" | "high" | "max";
 
 export type PromptCacheUsage = Readonly<{
   readTokens: number;
@@ -100,6 +102,8 @@ export function agentGenerationSettings(
     answerTokens?: number;
     thinkingTokens?: number;
     reasoningEffort?: ReasoningEffort;
+    /** Written beside `thinking`, not inside it, so compatible gateways emit `reasoning_effort`. */
+    providerReasoningEffort?: ProviderReasoningEffort;
   } = {},
 ) {
   const thinkingMode: ThinkingMode = options.thinking ?? "disabled";
@@ -110,6 +114,9 @@ export function agentGenerationSettings(
         ? { reasoningEffort: options.reasoningEffort }
         : {}),
     },
+    ...(options.providerReasoningEffort
+      ? { reasoningEffort: options.providerReasoningEffort }
+      : {}),
   };
   const providerId = typeof model === "string"
     ? model
diff --git a/frontend/src/mastra/consultation-tools.ts b/frontend/src/mastra/consultation-tools.ts
index 47d9e899..c5c3fd8f 100644
--- a/frontend/src/mastra/consultation-tools.ts
+++ b/frontend/src/mastra/consultation-tools.ts
@@ -25,7 +25,7 @@ import { createConsultationPlan, type ConsultationPlan } from "../lib/consultati
 import type { TechniqueAuditRow, WorkflowReceipt } from "../lib/consultation-agent-events.ts";
 import { normalizeTechniqueAuditRows } from "../lib/consultation-technique-audit.ts";
 import type { AgentModelFinishReason, EvidenceCardTelemetry } from "../lib/agent-observability.ts";
-import { agentGenerationSettings, AGENT_SLICE_ANSWER_OUTPUT_TOKENS, AGENT_SLICE_THINKING_OUTPUT_TOKENS } from "../lib/agent-generation-settings.ts";
+import { agentGenerationSettings, AGENT_SLICE_ANSWER_OUTPUT_TOKENS, AGENT_SLICE_THINKING_OUTPUT_TOKENS, type ProviderReasoningEffort } from "../lib/agent-generation-settings.ts";
 import { chartCalculationProgressLabel, evidenceLookupActivityLabel } from "../lib/consultation-activity-labels.ts";
 import { consultDomainConcurrency } from "../lib/consult/prep-flags.ts";
 import {
@@ -176,6 +176,45 @@ export const CONSULTATION_EVIDENCE_LOOKUP_TOOL_ID = "read-consultation-evidence"
 export const MAX_EVIDENCE_LOOKUPS_PER_TURN = 3;
 export const CONSULTATION_WINDOW_CALC_TOOL_ID = "run-jyotish-window-consultation";
 
+/**
+ * Answer-step reasoning effort. Unset or blank keeps today's request body.
+ * An illegal value warns once and is treated as unset.
+ */
+const CONSULT_ANSWER_REASONING_EFFORTS = ["low", "medium", "high", "max"] as const;
+let consultAnswerReasoningEffortWarned = false;
+
+export function consultAnswerReasoningEffort(): ProviderReasoningEffort | undefined {
+  const raw = process.env.CONSULT_ANSWER_REASONING_EFFORT;
+  if (raw === undefined || raw.trim() === "") return undefined;
+  const value = raw.trim();
+  if ((CONSULT_ANSWER_REASONING_EFFORTS as readonly string[]).includes(value)) {
+    return value as ProviderReasoningEffort;
+  }
+  if (!consultAnswerReasoningEffortWarned) {
+    consultAnswerReasoningEffortWarned = true;
+    console.warn(`CONSULT_ANSWER_REASONING_EFFORT=${value} is not low, medium, high, or max; leaving reasoning effort unset.`);
+  }
+  return undefined;
+}
+
+type AnswerEffortInput = {
+  answerReady?: boolean;
+  answerFromStart?: boolean;
+  model?: unknown;
+};
+
+function answerEffortOptions(input: AnswerEffortInput) {
+  const effort = consultAnswerReasoningEffort();
+  const answerStep = input.answerFromStart === true || input.answerReady === true;
+  if (!effort || !answerStep) return {};
+  return {
+    providerOptions: agentGenerationSettings(input.model, {
+      thinking: "enabled",
+      providerReasoningEffort: effort,
+    }).providerOptions,
+  };
+}
+
 /**
  * Step 0 narrows with activeTools and leaves toolChoice auto. Thinking-mode
  * providers reject named and required toolChoice (BUG-282, BUG-937).
@@ -185,35 +224,77 @@ export const CONSULTATION_WINDOW_CALC_TOOL_ID = "run-jyotish-window-consultation
  * Later steps stay auto.
  *
  * Window and general agents must not share the natal hook.
+ *
+ * The answer step is any step that already holds a calculation (`answerReady`,
+ * the same flag as the answer clock) or a no-tool public-daily turn
+ * (`answerFromStart`). A retry starts a new stream at step 0 with the
+ * calculation already in hand, so the step number is not part of the check.
+ * The main loop's step 0 still has `answerReady` false.
  */
-export function consultationNatalPrepareStep(input: { stepNumber: number; rectificationOffer?: boolean }) {
-  if (input.stepNumber !== 0) return { toolChoice: "auto" as const };
+type ConsultPrepareInput = {
+  rectificationOffer?: boolean;
+  answerReady?: boolean;
+  answerFromStart?: boolean;
+  model?: unknown;
+};
+
+export function consultationNatalPrepareStep(input: { stepNumber: 0 } & ConsultPrepareInput): {
+  activeTools: string[];
+  toolChoice: "auto";
+};
+export function consultationNatalPrepareStep(input: { stepNumber: number } & ConsultPrepareInput): {
+  toolChoice: "auto";
+  activeTools?: string[];
+};
+export function consultationNatalPrepareStep(input: { stepNumber: number } & ConsultPrepareInput) {
+  const effort = answerEffortOptions(input);
+  if (input.stepNumber !== 0) return { toolChoice: "auto" as const, ...effort };
   if (!input.rectificationOffer) {
     return {
       activeTools: [CONSULTATION_NATAL_CALC_TOOL_ID],
       toolChoice: "auto" as const,
+      ...effort,
     };
   }
   return {
     activeTools: [CONSULTATION_NATAL_CALC_TOOL_ID, RECTIFICATION_OFFER_TOOL_ID],
     toolChoice: "auto" as const,
+    ...effort,
   };
 }
 
-export function consultationWindowPrepareStep(input: { stepNumber: number; rectificationOffer?: boolean }) {
-  if (input.stepNumber !== 0) return { toolChoice: "auto" as const };
+export function consultationWindowPrepareStep(input: {
+  stepNumber: number;
+  rectificationOffer?: boolean;
+  answerReady?: boolean;
+  answerFromStart?: boolean;
+  model?: unknown;
+}) {
+  const effort = answerEffortOptions(input);
+  if (input.stepNumber !== 0) return { toolChoice: "auto" as const, ...effort };
   if (!input.rectificationOffer) {
     return {
       activeTools: [CONSULTATION_WINDOW_CALC_TOOL_ID],
       toolChoice: "auto" as const,
+      ...effort,
     };
   }
   return {
     activeTools: [CONSULTATION_WINDOW_CALC_TOOL_ID, RECTIFICATION_OFFER_TOOL_ID],
     toolChoice: "auto" as const,
+    ...effort,
   };
 }
 
+export function consultAnswerEffortProviderOptions(model?: unknown) {
+  return answerEffortOptions({ answerReady: true, model });
+}
+
+/** Public daily has no chart tool. Its first step is already the answer. */
+export function consultationGeneralPrepareStep(input: { stepNumber: number; model?: unknown }) {
+  return answerEffortOptions({ ...input, answerFromStart: true });
+}
+
 const CONSULTATION_DOMAIN_DURATION_MS = 31_000;
 const CONSULTATION_ANSWER_RESERVE_MS = 45_000;
 export const CONSULTATION_DOMAIN_WALL_CLOCK_MS = AGENT_TIMEOUT_MS - CONSULTATION_ANSWER_RESERVE_MS;
diff --git a/frontend/tests/answer-effort-probe.mts b/frontend/tests/answer-effort-probe.mts
new file mode 100644
index 00000000..0b008c25
--- /dev/null
+++ b/frontend/tests/answer-effort-probe.mts
@@ -0,0 +1,203 @@
+/**
+ * Request-body probe for issue #65.
+ *
+ * --dry-run needs no model key. It sends one agent.stream per shape through
+ * Mastra's OpenAI-compatible channel and prints the thinking / reasoning_effort
+ * / max_tokens fields of each request. A fake fetch never leaves the machine.
+ *
+ * Shapes:
+ *   (a) consultationGenerationSettings (today's answer settings)
+ *   (b) consultationSliceGenerationSettings (nested reasoningEffort, no caller)
+ *   (c) providerReasoningEffort low / medium / high / max at the top level
+ *
+ * With DS_KEY or BACKTEST_MODEL_KEY, the same file can ask the provider which
+ * efforts it accepts. This checkout does not run that half.
+ */
+import { writeFileSync } from "node:fs";
+
+import { Agent } from "@mastra/core/agent";
+import { createTool } from "@mastra/core/tools";
+import { z } from "zod";
+
+import { agentGenerationSettings } from "../src/lib/agent-generation-settings.ts";
+import {
+  consultationContinueGenerationSettings,
+  consultationGenerationSettings,
+  consultationNatalPrepareStep,
+  consultationSliceGenerationSettings,
+} from "../src/mastra/consultation-tools.ts";
+
+const MODEL = {
+  providerId: "deepseek",
+  modelId: "deepseek-flash",
+  url: "http://127.0.0.1:9/v1",
+  apiKey: "dummy",
+};
+
+const markTool = createTool({
+  id: "run-jyotish-consultation",
+  description: "Mark the calculation done.",
+  inputSchema: z.object({}).strict(),
+  execute: async () => ({ ok: true }),
+});
+
+function sse(chunks: unknown[]) {
+  const body = `${chunks.map((chunk) => `data: ${JSON.stringify(chunk)}\n\n`).join("")}data: [DONE]\n\n`;
+  return new Response(body, {
+    status: 200,
+    headers: { "content-type": "text/event-stream" },
+  });
+}
+
+function toolCallChunk() {
+  return sse([
+    {
+      id: "chatcmpl-tool",
+      object: "chat.completion.chunk",
+      choices: [{
+        index: 0,
+        delta: {
+          role: "assistant",
+          tool_calls: [{
+            index: 0,
+            id: "call_1",
+            type: "function",
+            function: { name: "run-jyotish-consultation", arguments: "" },
+          }],
+        },
+        finish_reason: null,
+      }],
+    },
+    {
+      id: "chatcmpl-tool",
+      object: "chat.completion.chunk",
+      choices: [{
+        index: 0,
+        delta: { tool_calls: [{ index: 0, function: { arguments: "{}" } }] },
+        finish_reason: null,
+      }],
+    },
+    {
+      id: "chatcmpl-tool",
+      object: "chat.completion.chunk",
+      choices: [{ index: 0, delta: {}, finish_reason: "tool_calls" }],
+    },
+  ]);
+}
+
+function textChunk() {
+  return sse([
+    {
+      id: "chatcmpl-text",
+      object: "chat.completion.chunk",
+      choices: [{ index: 0, delta: { role: "assistant", content: "ok" }, finish_reason: null }],
+    },
+    {
+      id: "chatcmpl-text",
+      object: "chat.completion.chunk",
+      choices: [{ index: 0, delta: {}, finish_reason: "stop" }],
+      usage: { prompt_tokens: 1, completion_tokens: 1, completion_tokens_details: { reasoning_tokens: 0 } },
+    },
+  ]);
+}
+
+export function requestFields(body: unknown) {
+  const record = body && typeof body === "object" ? body as Record<string, unknown> : {};
+  return {
+    thinking: record.thinking ?? null,
+    reasoning_effort: record.reasoning_effort ?? null,
+    max_tokens: record.max_tokens ?? null,
+  };
+}
+
+export async function captureStreamBodies(input: {
+  settings: ReturnType<typeof consultationGenerationSettings>;
+  prepareStep?: (step: { stepNumber: number }) => { toolChoice?: "auto"; activeTools?: string[]; providerOptions?: unknown };
+  steps: 1 | 2;
+}): Promise<unknown[]> {
+  const bodies: unknown[] = [];
+  const previous = globalThis.fetch;
+  globalThis.fetch = async (_url, init) => {
+    const raw = init?.body;
+    bodies.push(JSON.parse(typeof raw === "string" ? raw : "{}"));
+    return bodies.length === 1 && input.steps === 2 ? toolCallChunk() : textChunk();
+  };
+  try {
+    const agent = new Agent({
+      id: `answer-effort-probe-${bodies.length}-${Date.now()}`,
+      name: "answer-effort-probe",
+      instructions: "Reply ok.",
+      model: MODEL,
+      tools: { "run-jyotish-consultation": markTool },
+    });
+    const result = await agent.stream([{ role: "user", content: "ping" }], {
+      maxSteps: input.steps,
+      ...input.settings,
+      ...(input.prepareStep ? { prepareStep: input.prepareStep } : {}),
+    } as never);
+    const drained: unknown[] = [];
+    for await (const chunk of result.fullStream) drained.push(chunk);
+    if (drained.length === 0) throw new Error("probe stream produced no chunks");
+  } finally {
+    globalThis.fetch = previous;
+  }
+  return bodies;
+}
+
+async function main() {
+  if (!process.argv.includes("--dry-run")) {
+    console.log("A/B 未跑：缺模型 key。只做请求体拦截时加上 --dry-run。");
+    return;
+  }
+  const model = MODEL;
+  const natal = await captureStreamBodies({
+    settings: consultationGenerationSettings(model),
+    prepareStep: (input) => consultationNatalPrepareStep({
+      stepNumber: input.stepNumber,
+      answerReady: input.stepNumber >= 1,
+    }),
+    steps: 2,
+  });
+  const slice = await captureStreamBodies({
+    settings: consultationSliceGenerationSettings(model),
+    steps: 1,
+  });
+  const topLevel = [];
+  for (const effort of ["low", "medium", "high", "max"] as const) {
+    const bodies = await captureStreamBodies({
+      settings: agentGenerationSettings(model, { thinking: "enabled", providerReasoningEffort: effort }),
+      steps: 1,
+    });
+    topLevel.push({ effort, fields: requestFields(bodies[0]) });
+  }
+  const continued = await captureStreamBodies({
+    settings: consultationContinueGenerationSettings(model),
+    steps: 1,
+  });
+  const report = {
+    fixtureProvenance: "frontend/tests/answer-effort-probe.mts --dry-run",
+    a_consultationGenerationSettings: natal.map(requestFields),
+    b_consultationSliceGenerationSettings: slice.map(requestFields),
+    c_providerReasoningEffort: topLevel,
+    continue: continued.map(requestFields),
+  };
+  console.log(JSON.stringify(report, null, 2));
+  if (process.argv.includes("--save-baseline")) {
+    const file = new URL("./fixtures/answer-effort-switch-20261013.json", import.meta.url);
+    writeFileSync(file, `${JSON.stringify({
+      fixtureProvenance: "node node_modules/tsx/dist/cli.mjs tests/answer-effort-probe.mts --dry-run --save-baseline（开关尚未读取环境变量时的整份请求体）",
+      natal: natal,
+      continue: continued,
+    }, null, 2)}\n`);
+  }
+}
+
+const invoked = process.argv[1]
+  ? new URL(`file:///${process.argv[1].replace(/\\/g, "/")}`).href === import.meta.url
+  : false;
+if (invoked) {
+  main().catch((error) => {
+    console.error(error);
+    process.exitCode = 1;
+  });
+}
diff --git a/frontend/tests/answer-effort-switch-20261013.test.ts b/frontend/tests/answer-effort-switch-20261013.test.ts
new file mode 100644
index 00000000..20589d37
--- /dev/null
+++ b/frontend/tests/answer-effort-switch-20261013.test.ts
@@ -0,0 +1,111 @@
+/**
+ * Answer-step reasoning effort (issue #65).
+ * The unset bodies were captured before this switch read the environment:
+ * node node_modules/tsx/dist/cli.mjs tests/answer-effort-probe.mts --dry-run --save-baseline
+ */
+import assert from "node:assert/strict";
+import { readFileSync } from "node:fs";
+import test from "node:test";
+
+import {
+  consultAnswerReasoningEffort,
+  consultationContinueGenerationSettings,
+  consultationGeneralPrepareStep,
+  consultationGenerationSettings,
+  consultationNatalPrepareStep,
+} from "../src/mastra/consultation-tools.ts";
+import { captureStreamBodies } from "./answer-effort-probe.mts";
+
+const MODEL = {
+  providerId: "deepseek",
+  modelId: "deepseek-flash",
+  url: "http://127.0.0.1:9/v1",
+  apiKey: "dummy",
+};
+
+const fixture = JSON.parse(readFileSync(new URL("./fixtures/answer-effort-switch-20261013.json", import.meta.url), "utf8")) as {
+  natal: unknown[];
+  continue: unknown[];
+};
+
+function withoutEffort(body: unknown) {
+  const copy = { ...(body as Record<string, unknown>) };
+  delete copy.reasoning_effort;
+  return copy;
+}
+
+test("an illegal effort warns once and stays unset", () => {
+  const previous = process.env.CONSULT_ANSWER_REASONING_EFFORT;
+  const seen: unknown[][] = [];
+  const warn = console.warn;
+  console.warn = (...args: unknown[]) => { seen.push(args); };
+  process.env.CONSULT_ANSWER_REASONING_EFFORT = "turbo";
+  try {
+    assert.equal(consultAnswerReasoningEffort(), undefined);
+    assert.equal(consultAnswerReasoningEffort(), undefined);
+    assert.equal(seen.length, 1);
+  } finally {
+    console.warn = warn;
+    if (previous === undefined) delete process.env.CONSULT_ANSWER_REASONING_EFFORT;
+    else process.env.CONSULT_ANSWER_REASONING_EFFORT = previous;
+  }
+});
+
+test("unset effort keeps the pre-change request bodies", async () => {
+  const previous = process.env.CONSULT_ANSWER_REASONING_EFFORT;
+  delete process.env.CONSULT_ANSWER_REASONING_EFFORT;
+  try {
+    const natal = await captureStreamBodies({
+      settings: consultationGenerationSettings(MODEL),
+      prepareStep: (input) => consultationNatalPrepareStep({
+        stepNumber: input.stepNumber,
+        answerReady: input.stepNumber >= 1,
+        model: MODEL,
+      }),
+      steps: 2,
+    });
+    const continued = await captureStreamBodies({
+      settings: consultationContinueGenerationSettings(MODEL),
+      steps: 1,
+    });
+    assert.deepEqual(natal, fixture.natal);
+    assert.deepEqual(continued, fixture.continue);
+  } finally {
+    if (previous === undefined) delete process.env.CONSULT_ANSWER_REASONING_EFFORT;
+    else process.env.CONSULT_ANSWER_REASONING_EFFORT = previous;
+  }
+});
+
+test("low effort is only on the answer step", async () => {
+  const previous = process.env.CONSULT_ANSWER_REASONING_EFFORT;
+  process.env.CONSULT_ANSWER_REASONING_EFFORT = "low";
+  try {
+    const natal = await captureStreamBodies({
+      settings: consultationGenerationSettings(MODEL),
+      prepareStep: (input) => consultationNatalPrepareStep({
+        stepNumber: input.stepNumber,
+        answerReady: input.stepNumber >= 1,
+        model: MODEL,
+      }),
+      steps: 2,
+    }) as Record<string, unknown>[];
+    assert.equal(natal[0]?.reasoning_effort, undefined);
+    assert.deepEqual(natal[0], fixture.natal[0]);
+    assert.equal(natal[1]?.reasoning_effort, "low");
+    assert.deepEqual(withoutEffort(natal[1]), fixture.natal[1]);
+    const continued = await captureStreamBodies({
+      settings: consultationContinueGenerationSettings(MODEL),
+      steps: 1,
+    });
+    assert.deepEqual(continued, fixture.continue);
+    const daily = await captureStreamBodies({
+      settings: consultationGenerationSettings(MODEL),
+      prepareStep: (input) => consultationGeneralPrepareStep({ stepNumber: input.stepNumber, model: MODEL }),
+      steps: 1,
+    }) as Record<string, unknown>[];
+    assert.equal(daily[0]?.reasoning_effort, "low");
+  } finally {
+    if (previous === undefined) delete process.env.CONSULT_ANSWER_REASONING_EFFORT;
+    else process.env.CONSULT_ANSWER_REASONING_EFFORT = previous;
+  }
+});
diff --git a/frontend/tests/chart-request-skip-contract.test.ts b/frontend/tests/chart-request-skip-contract.test.ts
new file mode 100644
index 00000000..b24c519d
--- /dev/null
+++ b/frontend/tests/chart-request-skip-contract.test.ts
@@ -0,0 +1,44 @@
+import assert from "node:assert/strict";
+import { readdirSync, readFileSync, statSync } from "node:fs";
+import path from "node:path";
+import { fileURLToPath } from "node:url";
+import test from "node:test";
+
+const frontendRoot = fileURLToPath(new URL("../", import.meta.url));
+const roots = ["src/app/api", "src/lib"];
+// Empty on purpose. A call site that cannot send skip must be listed here with a reason.
+const allowlist: string[] = [];
+
+function walk(dir: string): string[] {
+  const found: string[] = [];
+  for (const name of readdirSync(dir)) {
+    const full = path.join(dir, name);
+    if (statSync(full).isDirectory()) found.push(...walk(full));
+    else if (name.endsWith(".ts") || name.endsWith(".tsx")) found.push(full);
+  }
+  return found;
+}
+
+test("python /api/chart bodies from app/api and lib send skip", () => {
+  const failures: string[] = [];
+  for (const root of roots) {
+    for (const file of walk(path.join(frontendRoot, root))) {
+      const rel = path.relative(frontendRoot, file).replaceAll("\\", "/");
+      if (allowlist.includes(rel)) continue;
+      const text = readFileSync(file, "utf8");
+      const re = /["']\/api\/chart["']/g;
+      let match: RegExpExecArray | null;
+      while ((match = re.exec(text))) {
+        const idx = match.index;
+        const before = text.slice(Math.max(0, idx - 120), idx);
+        if (!/postPython\s*\(|callEngine\s*\(|fetchEngine\s*\(|fetch\s*\(/.test(before)) continue;
+        const window = text.slice(Math.max(0, idx - 40), idx + 420);
+        const sendsSkip = window.includes("skip_vedastro_main_entry_overview: true")
+          || window.includes("transitChartBody(")
+          || (/["']\/api\/chart["']\s*,\s*payload/.test(window) && text.includes("skip_vedastro_main_entry_overview: true"));
+        if (!sendsSkip) failures.push(`${rel}:${text.slice(0, idx).split("\n").length}`);
+      }
+    }
+  }
+  assert.deepEqual(failures, []);
+});
diff --git a/frontend/tests/consult-answer-clock-20261001.test.ts b/frontend/tests/consult-answer-clock-20261001.test.ts
index 4e567a49..8cac8b56 100644
--- a/frontend/tests/consult-answer-clock-20261001.test.ts
+++ b/frontend/tests/consult-answer-clock-20261001.test.ts
@@ -50,6 +50,9 @@ test("the consult route starts the general / no-birth-minute loop on the answer
   );
   // That branch still streams on the shared loop signal and keeps its traces.
   const general = route.match(/if \(usesPublicDailyGeneralAgent\(consultationMode, generalDailyContext\)\) \{[\s\S]*?\n {4}\}\n/)?.[0] ?? "";
-  assert.match(general, /await streamWithOverflowRetry\(agent\)/);
+  // 原值：await streamWithOverflowRetry(agent)
+  // 新值：await streamWithOverflowRetry(agent, generalStreamOptions)
+  // 原因：公开日历从第一步就是答题。开关打开时这一步要带推理强度；不传选项就没有 prepareStep，强度加不上去。开关不设时这些选项不改请求体。
+  assert.match(general, /await streamWithOverflowRetry\(agent, generalStreamOptions\)/);
   assert.match(general, /onError: \(error\) => settleRun\(\s*cancel,/);
 });
diff --git a/frontend/tests/fixtures/answer-effort-switch-20261013.json b/frontend/tests/fixtures/answer-effort-switch-20261013.json
new file mode 100644
index 00000000..a78368a3
--- /dev/null
+++ b/frontend/tests/fixtures/answer-effort-switch-20261013.json
@@ -0,0 +1,131 @@
+{
+  "fixtureProvenance": "node node_modules/tsx/dist/cli.mjs scripts/research/answer-effort-probe.mts --dry-run --save-baseline（开关尚未读取环境变量时的整份请求体）",
+  "natal": [
+    {
+      "model": "deepseek-flash",
+      "max_tokens": 24576,
+      "thinking": {
+        "type": "enabled"
+      },
+      "messages": [
+        {
+          "role": "system",
+          "content": "Reply ok."
+        },
+        {
+          "role": "user",
+          "content": "ping"
+        }
+      ],
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "run-jyotish-consultation",
+            "description": "Mark the calculation done.",
+            "parameters": {
+              "type": "object",
+              "properties": {},
+              "additionalProperties": false,
+              "$schema": "http://json-schema.org/draft-07/schema#",
+              "required": []
+            }
+          }
+        }
+      ],
+      "tool_choice": "auto",
+      "stream": true
+    },
+    {
+      "model": "deepseek-flash",
+      "max_tokens": 24576,
+      "thinking": {
+        "type": "enabled"
+      },
+      "messages": [
+        {
+          "role": "system",
+          "content": "Reply ok."
+        },
+        {
+          "role": "user",
+          "content": "ping"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "id": "call_1",
+              "type": "function",
+              "function": {
+                "name": "run-jyotish-consultation",
+                "arguments": "{}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_1",
+          "content": "{\"ok\":true}"
+        }
+      ],
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "run-jyotish-consultation",
+            "description": "Mark the calculation done.",
+            "parameters": {
+              "type": "object",
+              "properties": {},
+              "additionalProperties": false,
+              "$schema": "http://json-schema.org/draft-07/schema#",
+              "required": []
+            }
+          }
+        }
+      ],
+      "tool_choice": "auto",
+      "stream": true
+    }
+  ],
+  "continue": [
+    {
+      "model": "deepseek-flash",
+      "max_tokens": 16384,
+      "thinking": {
+        "type": "disabled"
+      },
+      "messages": [
+        {
+          "role": "system",
+          "content": "Reply ok."
+        },
+        {
+          "role": "user",
+          "content": "ping"
+        }
+      ],
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "run-jyotish-consultation",
+            "description": "Mark the calculation done.",
+            "parameters": {
+              "type": "object",
+              "properties": {},
+              "additionalProperties": false,
+              "$schema": "http://json-schema.org/draft-07/schema#",
+              "required": []
+            }
+          }
+        }
+      ],
+      "tool_choice": "auto",
+      "stream": true
+    }
+  ]
+}
diff --git a/frontend/tests/synastry-chart-skip.test.ts b/frontend/tests/synastry-chart-skip.test.ts
new file mode 100644
index 00000000..ffc03949
--- /dev/null
+++ b/frontend/tests/synastry-chart-skip.test.ts
@@ -0,0 +1,103 @@
+import assert from "node:assert/strict";
+import { spawnSync } from "node:child_process";
+import { fileURLToPath } from "node:url";
+import test from "node:test";
+
+const frontendRoot = fileURLToPath(new URL("../", import.meta.url));
+const partnerId = "11111111-1111-4111-8111-111111111111";
+
+function runRoute(script: string): { romance: Record<string, unknown>; business: Record<string, unknown>; maxChart: number; maxVarga: number; chartSkip: boolean } {
+  const result = spawnSync(process.execPath, [
+    "--experimental-test-module-mocks",
+    "--import",
+    "tsx",
+    "--input-type=module",
+    "--eval",
+    script,
+  ], { cwd: frontendRoot, encoding: "utf8", timeout: 20000 });
+  assert.equal(result.status, 0, result.stderr || result.stdout);
+  const line = result.stdout.trim().split("\n").at(-1) ?? "{}";
+  return JSON.parse(line);
+}
+
+test("synastry charts send skip concurrently and keep the stubbed JSON", () => {
+  const script = String.raw`
+    import { mock } from "node:test";
+    import { pathToFileURL } from "node:url";
+    const moduleUrl = (path) => pathToFileURL(process.cwd() + "/" + path).href;
+    mock.module("server-only", { namedExports: {} });
+    mock.module("@/lib/account-deletion-server", { namedExports: { refuseWhenAccountDeletionPending: async () => null } });
+    mock.module("@/lib/product-access", { namedExports: { isProductEnabled: async () => true } });
+    mock.module("@/lib/request-rate-limit", { namedExports: { consumeUserRequestRateLimit: () => ({ ok: true, retryAfterSeconds: 1 }) } });
+    mock.module("@/lib/supabase/server", { namedExports: { createServerSupabaseClient: async () => ({ auth: { getUser: async () => ({ data: { user: { id: "user-1" } }, error: null }) } }) } });
+    mock.module("@/lib/subject-birth", { namedExports: {
+      ConsultationSubjectError: class ConsultationSubjectError extends Error {},
+      loadSubjectBirth: async () => ({ chartable: true, row: { name: "fixture" } }),
+    } });
+    mock.module("@/lib/server-owned-birth-profile", { namedExports: { globalBirthProfileFromAccountRow: (row) => row } });
+    mock.module("@/lib/global-birth-payloads", { namedExports: {
+      synastryBirthPayload: async () => ({ year: 1990, month: 1, day: 1, hour: 12, minute: 0, lat: 28.6139, lon: 77.209, tz: 5.5, ayanamsa: "lahiri" }),
+    } });
+    const chartBodies = [];
+    const maxInflight = { chart: 0, varga: 0 };
+    const inflight = { chart: 0, varga: 0 };
+    const waiters = { chart: [], varga: [] };
+    globalThis.fetch = async (inputUrl, init) => {
+      const url = String(inputUrl);
+      const body = JSON.parse(String(init.body));
+      const kind = url.endsWith("/api/chart") ? "chart" : url.endsWith("/api/varga_full") ? "varga" : "other";
+      if (kind === "chart") chartBodies.push(body);
+      if (kind === "chart" || kind === "varga") {
+        inflight[kind] += 1;
+        maxInflight[kind] = Math.max(maxInflight[kind], inflight[kind]);
+        await new Promise((resolve) => {
+          waiters[kind].push(resolve);
+          if (waiters[kind].length >= 2) {
+            const pending = waiters[kind].splice(0, 2);
+            pending.forEach((fn) => fn());
+          }
+        });
+        inflight[kind] -= 1;
+      }
+      if (url.endsWith("/api/chart")) {
+        return Response.json({
+          planets: { Moon: { lon: 12.5, sign: "Aries", nakshatra: "Ashwini", nakshatra_pada: 1, nakshatra_lord: "Ketu" } },
+          ascendant: { sign: "Leo" },
+        });
+      }
+      if (url.endsWith("/api/varga_full")) {
+        const divisions = body.divisions || [];
+        if (divisions.includes("D9")) {
+          return Response.json({ source: "engine", result: { D9_Navamsa: { ascendant: { sign: "Virgo" }, planets: { Moon: { sign: "Cancer" }, Venus: { sign: "Taurus" }, Mars: { sign: "Aries" } } } } });
+        }
+        return Response.json({ source: "engine", result: {
+          D2_Hora: { planets: { Moon: { sign: "Taurus" } } },
+          D10_Dasamsa: { ascendant: { sign: "Libra" } },
+          D11_Rudramsa: { planets: { Sun: { sign: "Capricorn" } } },
+        } });
+      }
+      if (url.endsWith("/api/synastry")) return Response.json({ total_score: 18, max_score: 36 });
+      return Response.json({}, { status: 500 });
+    };
+    const { POST } = await import(moduleUrl("src/app/api/synastry/route.ts"));
+    const romanceRes = await POST(new Request("https://example.test/api/synastry", { method: "POST", body: JSON.stringify({ partnerChartProfileId: "${partnerId}", relationshipType: "romance" }) }));
+    const romance = await romanceRes.json();
+    const businessRes = await POST(new Request("https://example.test/api/synastry", { method: "POST", body: JSON.stringify({ partnerChartProfileId: "${partnerId}", relationshipType: "business" }) }));
+    const business = await businessRes.json();
+    console.log(JSON.stringify({ romance, business, maxChart: maxInflight.chart, maxVarga: maxInflight.varga, chartSkip: chartBodies.every((body) => body.skip_vedastro_main_entry_overview === true) && chartBodies.length === 4 }));
+  `;
+  const got = runRoute(script);
+  assert.equal(got.maxChart, 2);
+  assert.equal(got.maxVarga, 2);
+  assert.equal(got.chartSkip, true);
+  assert.equal(got.romance.status, "ok");
+  assert.equal(got.romance.method, "ashtakoot_plus_moon_nakshatra_d9");
+  assert.deepEqual(got.romance.selfChart, {
+    moon: { sign: "Aries", nakshatra: "Ashwini", pada: 1, lord: "Ketu", longitude: 12.5 },
+    d9: { ascendant: { sign: "Virgo" }, moon: { sign: "Cancer" }, venus: { sign: "Taurus" }, mars: { sign: "Aries" }, source: "engine" },
+  });
+  assert.equal((got.romance.relationshipReport as { scoreBand?: string }).scoreBand, "mixed");
+  assert.equal(got.business.status, "ok");
+  assert.equal(got.business.method, "d2_d10_d11_business_screening_partial");
+  assert.equal((got.business.relationshipReport as { scoreBand?: string }).scoreBand, "not_scored");
+});
diff --git a/scripts/annual_pyjhora_replay.py b/scripts/annual_pyjhora_replay.py
index 01d9a4d4..5f02a453 100755
--- a/scripts/annual_pyjhora_replay.py
+++ b/scripts/annual_pyjhora_replay.py
@@ -9,6 +9,7 @@ from __future__ import annotations
 
 import contextlib
 import hashlib
+import importlib.util
 import json
 import subprocess
 import sys
@@ -22,6 +23,34 @@ from typing import Any, Callable
 
 ENGINE = "PyJHora/JHora"
 REPO_ROOT = Path(__file__).resolve().parents[1]
+_jhora_spec_cache: bool | None = None
+
+
+def reset_jhora_import_cache() -> None:
+    """Tests reset the process-local "is PyJHora installed?" answer."""
+    global _jhora_spec_cache
+    _jhora_spec_cache = None
+
+
+def _jhora_importable() -> bool:
+    global _jhora_spec_cache
+    if _jhora_spec_cache is None:
+        _jhora_spec_cache = importlib.util.find_spec("jhora") is not None
+    return _jhora_spec_cache
+
+
+def _no_jhora_tajaka_replay(settings: dict[str, Any]) -> dict[str, Any]:
+    """Same dict the child returns when ``import jhora`` raises ModuleNotFoundError."""
+    copied = dict(settings)
+    return {
+        "status": "blocked",
+        "reason": "pyjhora_ayanamsa_configuration_failed:ModuleNotFoundError",
+        "settings": copied,
+        "node_mode": {
+            "requested": copied.get("node_mode"),
+            "status": "unsupported_by_replay_adapter",
+        },
+    }
 _TAJAKA_YEAR_LORD_MODULE = "jhora.horoscope.transit.tajaka"
 _YEAR_LORD_MODULES = (
     "jhora.horoscope.dhasa.annual.mudda",
@@ -139,6 +168,10 @@ def _run_isolated_tajaka_year_lord(
         "settings": settings,
     }
     request_hash = _request_hash(request)
+    if not _jhora_importable():
+        replay = _no_jhora_tajaka_replay(settings)
+        replay["request_hash"] = request_hash
+        return replay
     command = [
         sys.executable,
         "-c",
@@ -276,6 +309,10 @@ def _run_isolated_annual_replay(
         "settings": settings,
     }
     request_hash = _request_hash(request)
+    if not _jhora_importable():
+        replay = _no_jhora_annual_replay(settings)
+        replay["request_hash"] = request_hash
+        return replay
     command = [
         sys.executable,
         "-c",
@@ -391,6 +428,18 @@ def _annual_worker_result(
     }
 
 
+def _no_jhora_annual_replay(settings: dict[str, Any]) -> dict[str, Any]:
+    """Same dict the annual child returns when ``import jhora`` raises ModuleNotFoundError."""
+    copied = dict(settings)
+    node_mode = {"requested": copied.get("node_mode"), "status": "unsupported_by_replay_adapter"}
+    reason = "pyjhora_annual_adapter_setup_failed:ModuleNotFoundError"
+    return _annual_worker_result(
+        {key: {"status": "blocked", "reason": reason} for key in _ANNUAL_SURFACE_KEYS},
+        settings=copied,
+        node_mode=node_mode,
+    )
+
+
 def _annual_replay_blocked(reason: str, request_hash: str, settings: dict[str, Any]) -> dict[str, Any]:
     return _annual_worker_result(
         {key: {"status": "blocked", "reason": reason} for key in _ANNUAL_SURFACE_KEYS},
diff --git a/scripts/api_heavy_compute_gate.py b/scripts/api_heavy_compute_gate.py
index cf9f250f..0067a9b9 100644
--- a/scripts/api_heavy_compute_gate.py
+++ b/scripts/api_heavy_compute_gate.py
@@ -8,13 +8,17 @@ must not call this gate.
 
 from __future__ import annotations
 
+import logging
 import os
 import threading
+import time
 
 DEFAULT_CONCURRENCY = 2
 DEFAULT_RETRY_AFTER_SECONDS = 2
 ENV_CONCURRENCY = "JYOTISH_HEAVY_COMPUTE_CONCURRENCY"
 ENV_RETRY_AFTER = "JYOTISH_HEAVY_COMPUTE_RETRY_AFTER_SECONDS"
+ENV_GATE_CHART_OVERVIEW = "JYOTISH_GATE_CHART_OVERVIEW"
+_LOG = logging.getLogger("jyotish.heavy_compute")
 
 HEAVY_COMPUTE_PATHS = frozenset(
     {
@@ -46,6 +50,26 @@ HEAVY_COMPUTE_PATHS = frozenset(
 _state_lock = threading.Lock()
 _semaphore: threading.BoundedSemaphore | None = None
 _retry_after_seconds = DEFAULT_RETRY_AFTER_SECONDS
+_limit = DEFAULT_CONCURRENCY
+_holders: list[dict] = []
+_holder_seq = 0
+
+
+class _HeldSlot:
+    """One acquired heavy-compute permit, paired with the path that holds it."""
+
+    def __init__(self, semaphore: threading.BoundedSemaphore, holder_id: int) -> None:
+        self._semaphore = semaphore
+        self._holder_id = holder_id
+        self._released = False
+
+    def release(self) -> None:
+        with _state_lock:
+            if self._released:
+                return
+            self._released = True
+            _holders[:] = [item for item in _holders if item["id"] != self._holder_id]
+            self._semaphore.release()
 
 
 class HeavyComputeBusy(RuntimeError):
@@ -80,7 +104,7 @@ def reset_heavy_compute_gate(
     retry_after_seconds: int | None = None,
 ) -> None:
     """Rebuild the process-wide semaphore. Tests must call this after env changes."""
-    global _semaphore, _retry_after_seconds
+    global _semaphore, _retry_after_seconds, _limit, _holders
     resolved_limit = (
         limit
         if limit is not None
@@ -93,31 +117,96 @@ def reset_heavy_compute_gate(
     )
     with _state_lock:
         _retry_after_seconds = resolved_retry
+        _limit = resolved_limit
+        _holders = []
         _semaphore = threading.BoundedSemaphore(resolved_limit)
 
 
 def _ensure_locked() -> tuple[threading.BoundedSemaphore, int]:
-    global _semaphore, _retry_after_seconds
+    global _semaphore, _retry_after_seconds, _limit
     if _semaphore is None:
-        limit = _parse_positive_int(os.environ.get(ENV_CONCURRENCY), DEFAULT_CONCURRENCY)
+        _limit = _parse_positive_int(os.environ.get(ENV_CONCURRENCY), DEFAULT_CONCURRENCY)
         _retry_after_seconds = _parse_positive_int(
             os.environ.get(ENV_RETRY_AFTER), DEFAULT_RETRY_AFTER_SECONDS
         )
-        _semaphore = threading.BoundedSemaphore(limit)
+        _semaphore = threading.BoundedSemaphore(_limit)
     return _semaphore, _retry_after_seconds
 
 
-def acquire_heavy_compute_slot(path: str) -> threading.BoundedSemaphore | None:
+def _holder_view(now: float) -> list[dict]:
+    return [
+        {
+            "path": item["path"],
+            "held_seconds": round(max(now - item["started"], 0.0), 3),
+        }
+        for item in _holders
+    ]
+
+
+def _acquire_recorded(path: str) -> _HeldSlot:
+    """Take one permit and remember which path holds it. Fail fast when full."""
+    global _holder_seq
+    with _state_lock:
+        semaphore, retry_after = _ensure_locked()
+        if not semaphore.acquire(blocking=False):
+            snapshot = _holder_view(time.monotonic())
+            failed = True
+            slot = None
+        else:
+            _holder_seq += 1
+            holder_id = _holder_seq
+            _holders.append({"id": holder_id, "path": path, "started": time.monotonic()})
+            failed = False
+            snapshot = []
+            slot = _HeldSlot(semaphore, holder_id)
+    if failed:
+        _LOG.warning(
+            "heavy_compute_busy rejected_path=%s holders=%s",
+            path,
+            snapshot,
+        )
+        raise HeavyComputeBusy(retry_after)
+    assert slot is not None
+    return slot
+
+
+def acquire_heavy_compute_slot(path: str) -> _HeldSlot | None:
     """Acquire a slot for a gated path. Light paths return None. Fail-fast on saturation."""
     if not is_heavy_compute_path(path):
         return None
+    return _acquire_recorded(path)
+
+
+def _chart_overview_gate_enabled() -> bool:
+    return os.environ.get(ENV_GATE_CHART_OVERVIEW, "").strip() == "1"
+
+
+def acquire_heavy_compute_slot_for_request(path: str, body: dict | None) -> _HeldSlot | None:
+    """Path gate, plus the chart-overview switch and the warm-consultation exception.
+
+    ``/api/chart`` stays out of ``HEAVY_COMPUTE_PATHS``. It takes a slot only when
+    ``JYOTISH_GATE_CHART_OVERVIEW=1``, the body does not skip the VedAstro overview,
+    and the body is not an async enqueue. Default is off.
+    """
+    payload = body if isinstance(body, dict) else {}
+    if path == "/api/consultation_workflow" and payload.get("consult_card_warm_only") is True:
+        return None
+    if path == "/api/chart" and _chart_overview_gate_enabled():
+        if payload.get("async") or payload.get("enqueue"):
+            return None
+        if payload.get("skip_vedastro_main_entry_overview"):
+            return None
+        return _acquire_recorded(path)
+    return acquire_heavy_compute_slot(path)
+
+
+def heavy_compute_status() -> dict:
+    """Process-local counts only. No request body and no user identity."""
     with _state_lock:
-        semaphore, retry_after = _ensure_locked()
-    if not semaphore.acquire(blocking=False):
-        raise HeavyComputeBusy(retry_after)
-    return semaphore
+        _ensure_locked()
+        return {"limit": _limit, "in_use": len(_holders)}
 
 
-def release_heavy_compute_slot(slot: threading.BoundedSemaphore | None) -> None:
+def release_heavy_compute_slot(slot: _HeldSlot | None) -> None:
     if slot is not None:
         slot.release()
diff --git a/scripts/jyotish_api_server.py b/scripts/jyotish_api_server.py
index 432467e5..79985813 100644
--- a/scripts/jyotish_api_server.py
+++ b/scripts/jyotish_api_server.py
@@ -122,13 +122,13 @@ except ModuleNotFoundError:  # pragma: no cover - script execution path
 try:
     from scripts.api_heavy_compute_gate import (
         HeavyComputeBusy,
-        acquire_heavy_compute_slot,
+        acquire_heavy_compute_slot_for_request,
         release_heavy_compute_slot,
     )
 except ModuleNotFoundError:  # pragma: no cover - script execution path
     from api_heavy_compute_gate import (
         HeavyComputeBusy,
-        acquire_heavy_compute_slot,
+        acquire_heavy_compute_slot_for_request,
         release_heavy_compute_slot,
     )
 
@@ -3357,14 +3357,10 @@ class JyotishAPIHandler(BaseHTTPRequestHandler, VedastroEvidenceMixin, SynastryM
         try:
             self._enforce_request_security(require_json=True)
             body = self._read_json_body()
-            # Chart-save warm builds today's packet on the same route. It must
-            # not take a heavy-compute slot: the cap is 2, and a question asked
-            # right after save would otherwise come back 429.
-            warm_only = (
-                path == '/api/consultation_workflow'
-                and body.get('consult_card_warm_only') is True
-            )
-            compute_slot = None if warm_only else acquire_heavy_compute_slot(path)
+            # Warm chart-save must not take a slot (BUG-1302). Chart overview
+            # gating is off unless JYOTISH_GATE_CHART_OVERVIEW=1. Both live in
+            # the gate module so this file only keeps the call.
+            compute_slot = acquire_heavy_compute_slot_for_request(path, body)
             if path == '/api/location/resolve':
                 city = str(body.get('city') or '').strip()
                 city_aliases = {'beijing': '北京', 'shanghai': '上海', 'guangzhou': '广州', 'shenzhen': '深圳'}
diff --git a/scripts/jyotish_engine.py b/scripts/jyotish_engine.py
index e2f14143..982f1077 100644
--- a/scripts/jyotish_engine.py
+++ b/scripts/jyotish_engine.py
@@ -10762,7 +10762,25 @@ def _planet_snapshot(planets, planet_name):
     }
 
 
+_oracle_progress_cache: dict = {}
+
+
+def reset_oracle_progress_snapshot_cache():
+    """Tests clear the static-file validation cache."""
+    _oracle_progress_cache.clear()
+
+
+def _oracle_progress_cache_key(oracle_file):
+    try:
+        stat = oracle_file.stat()
+    except OSError:
+        return None
+    mtime = getattr(stat, 'st_mtime_ns', stat.st_mtime)
+    return (str(oracle_file.resolve()), mtime, stat.st_size)
+
+
 def _oracle_progress_snapshot():
+    cache_key = None
     try:
         import json
         import subprocess
@@ -10772,6 +10790,9 @@ def _oracle_progress_snapshot():
 
         root = Path(__file__).resolve().parents[1]
         oracle_file = root / 'references' / 'oracle' / 'dasha_shadbala_oracle_cases.json'
+        cache_key = _oracle_progress_cache_key(oracle_file)
+        if cache_key is not None and cache_key in _oracle_progress_cache:
+            return _oracle_progress_cache[cache_key]
         with tempfile.TemporaryDirectory() as temp_dir:
             queue_path = Path(temp_dir) / 'oracle-queue.json'
             queue = subprocess.run(
@@ -10797,7 +10818,7 @@ def _oracle_progress_snapshot():
                 raise RuntimeError(validation.stderr.strip() or validation.stdout.strip())
             report = json.loads(validation.stdout)
         summary = report.get('summary', {})
-        return {
+        result = {
             'scope': 'external_oracle_evidence_validation',
             'collection_queue': 'external_oracle_collection_queue',
             'total_packets': summary.get('total_packets', 0),
@@ -10809,7 +10830,7 @@ def _oracle_progress_snapshot():
             'boundary': report.get('boundary', 'Dasha/Shadbala absolute values are not externally calibrated until enough packets pass validation.'),
         }
     except Exception as exc:
-        return {
+        result = {
             'scope': 'external_oracle_evidence_validation',
             'collection_queue': 'external_oracle_collection_queue',
             'total_packets': 0,
@@ -10820,6 +10841,9 @@ def _oracle_progress_snapshot():
             'promotion_rule': 'external_verified requires source_artifact, filled target values, and non-local-engine external evidence.',
             'boundary': f'Oracle progress unavailable: {exc}',
         }
+    if cache_key is not None:
+        _oracle_progress_cache[cache_key] = result
+    return result
 
 
 # Combustion orbs of the full reading (Step 5 shadbala block). The Moon, Rahu
diff --git a/scripts/research/api_chart_subprocess_ab.py b/scripts/research/api_chart_subprocess_ab.py
new file mode 100644
index 00000000..fccc7279
--- /dev/null
+++ b/scripts/research/api_chart_subprocess_ab.py
@@ -0,0 +1,291 @@
+"""A/B for issue #64. A is this branch with the PyJHora short-circuit forced off
+(the staging control flow: always spawn). B is the short-circuit on.
+
+Chart skip/concurrency is measured on the real chart call. The gate schemes
+O2/O3/O4 are simulated only and are not implemented.
+
+Volatile fields dropped before the report byte compare:
+generated_at, created_at, timestamp, elapsed_seconds, elapsed_ms, duration_ms, request_id.
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+import os
+import random
+import statistics
+import subprocess
+import sys
+import tempfile
+import threading
+import time
+from concurrent.futures import ThreadPoolExecutor
+from pathlib import Path
+
+ROOT = Path(__file__).resolve().parents[2]
+sys.path.insert(0, str(ROOT))
+sys.path.insert(0, str(ROOT / "scripts"))
+
+os.environ.setdefault("JYOTISH_API_CHART_CACHE_TTL_SECONDS", "0")
+os.environ.setdefault("VEDASTRO_ENABLE_NETWORK", "0")
+os.environ.setdefault("JYOTISH_GATE_CHART_OVERVIEW", "0")
+
+import annual_pyjhora_replay as replay  # noqa: E402
+import tajika as tajika_top  # noqa: E402
+from scripts.jyotish_api_server import JyotishAPIHandler  # noqa: E402
+from scripts.jyotish_engine import reset_oracle_progress_snapshot_cache  # noqa: E402
+from scripts.professional_report_reference import build_professional_report_reference  # noqa: E402
+from scripts.tajika import reset_external_pyjhora_python_cache  # noqa: E402
+
+BIRTH = {
+    "year": 1990,
+    "month": 1,
+    "day": 1,
+    "hour": 12,
+    "minute": 0,
+    "second": 0,
+    "lat": 28.6139,
+    "lon": 77.2090,
+    "tz": 5.5,
+    "ayanamsa": "lahiri",
+    "node_mode": "true",
+}
+PARTNER = {**BIRTH, "year": 1992, "day": 2}
+VOLATILE = {"generated_at", "created_at", "timestamp", "elapsed_seconds", "elapsed_ms", "duration_ms", "request_id"}
+
+
+class _Counter:
+    def __init__(self) -> None:
+        self.n = 0
+        self._real = subprocess.run
+        self._lock = threading.Lock()
+
+    def install(self) -> None:
+        def _run(*args, **kwargs):
+            with self._lock:
+                self.n += 1
+            return self._real(*args, **kwargs)
+
+        subprocess.run = _run  # type: ignore[assignment]
+        for module in list(sys.modules.values()):
+            bound = getattr(module, "subprocess", None)
+            if getattr(bound, "run", None) is self._real:
+                bound.run = _run  # type: ignore[union-attr]
+
+    def reset(self) -> None:
+        with self._lock:
+            self.n = 0
+
+
+def _median(values: list[float]) -> float:
+    return statistics.median(values) if values else 0.0
+
+
+def _strip(value):
+    if isinstance(value, dict):
+        return {key: _strip(item) for key, item in value.items() if key not in VOLATILE}
+    if isinstance(value, list):
+        return [_strip(item) for item in value]
+    return value
+
+
+def _canon(value) -> str:
+    return json.dumps(_strip(value), sort_keys=True, ensure_ascii=True, default=str)
+
+
+def _chart_summary(chart: dict) -> dict:
+    planets = chart.get("planets") if isinstance(chart, dict) else None
+    moon = planets.get("Moon") if isinstance(planets, dict) else None
+    lon = None
+    if isinstance(moon, dict):
+        lon = moon.get("lon", moon.get("longitude"))
+    return {
+        "ascendant": chart.get("ascendant") if isinstance(chart, dict) else None,
+        "planets": planets,
+        "moon_longitude": lon,
+    }
+
+
+def _one_chart(body: dict) -> tuple[float, dict]:
+    started = time.perf_counter()
+    chart = object.__new__(JyotishAPIHandler)._compute_chart_sync(body)
+    return time.perf_counter() - started, chart
+
+
+def _measure_charts(counter: _Counter) -> dict:
+    serial_times = []
+    concurrent_times = []
+    serial_subs = []
+    concurrent_subs = []
+    same = True
+    for _ in range(3):
+        counter.reset()
+        started = time.perf_counter()
+        _, self_a = _one_chart(BIRTH)
+        _, partner_a = _one_chart(PARTNER)
+        serial_times.append(time.perf_counter() - started)
+        serial_subs.append(counter.n)
+        counter.reset()
+        started = time.perf_counter()
+        with ThreadPoolExecutor(max_workers=2) as pool:
+            self_b = pool.submit(_one_chart, {**BIRTH, "skip_vedastro_main_entry_overview": True})
+            partner_b = pool.submit(_one_chart, {**PARTNER, "skip_vedastro_main_entry_overview": True})
+            _, self_chart = self_b.result()
+            _, partner_chart = partner_b.result()
+        concurrent_times.append(time.perf_counter() - started)
+        concurrent_subs.append(counter.n)
+        same = same and _canon(_chart_summary(self_a)) == _canon(_chart_summary(self_chart))
+        same = same and _canon(_chart_summary(partner_a)) == _canon(_chart_summary(partner_chart))
+    return {
+        "serial_seconds_median": round(_median(serial_times), 4),
+        "concurrent_skip_seconds_median": round(_median(concurrent_times), 4),
+        "serial_subprocesses_median": _median(serial_subs),
+        "concurrent_skip_subprocesses_median": _median(concurrent_subs),
+        "summary_bytes_match": same,
+    }
+
+
+def _measure_report(counter: _Counter, *, spawn: bool) -> dict:
+    times = []
+    subs = []
+    payload = None
+    body = {**BIRTH, "format": "markdown", "edition": "full_data", "languages": ["zh", "en"], "include_fact_tables": True}
+    for _ in range(3):
+        replay.reset_jhora_import_cache()
+        if spawn:
+            replay._jhora_spec_cache = True
+        reset_oracle_progress_snapshot_cache()
+        reset_external_pyjhora_python_cache()
+        tajika_top.reset_external_pyjhora_python_cache()
+        counter.reset()
+        started = time.perf_counter()
+        payload = build_professional_report_reference(
+            object.__new__(JyotishAPIHandler),
+            body,
+        )
+        times.append(time.perf_counter() - started)
+        subs.append(counter.n)
+    return {
+        "seconds_median": round(_median(times), 4),
+        "subprocesses_median": _median(subs),
+        "payload": payload,
+    }
+
+
+def _simulate() -> list[dict]:
+    """One fixed minute. O2 adds a second pool and is the only scheme that touches BUG-717."""
+    rng = random.Random(64)
+    arrivals = [("report", index * 30.0, 13.0) for index in range(2)]
+    arrivals += [("consult", rng.random() * 60.0, 0.2) for _ in range(20)]
+    arrivals += [("score", rng.random() * 60.0, 8.0) for _ in range(6)]
+    arrivals.sort(key=lambda item: item[1])
+
+    def run(mode: str) -> dict:
+        heavy: list[tuple[float, str]] = []
+        light: list[float] = []
+        rejected = {"consult": 0, "report": 0, "score": 0}
+        accepted = {"consult": 0, "report": 0, "score": 0}
+        peak = 0
+
+        def occupy(at: float) -> tuple[int, int]:
+            running = [item for item in heavy if item[0] > at]
+            lights = [item for item in light if item > at]
+            return len(running), len(lights)
+
+        for kind, arrival, duration in arrivals:
+            heavy_n, light_n = occupy(arrival)
+            heavy = [item for item in heavy if item[0] > arrival]
+            light = [item for item in light if item > arrival]
+            placed = False
+            stamp = arrival
+            if mode == "O2" and kind == "consult":
+                if light_n < 2:
+                    light.append(arrival + duration)
+                    placed = True
+            elif mode == "O3" and kind == "consult":
+                start = arrival
+                while start <= arrival + 1.5:
+                    heavy_n, _light_n = occupy(start)
+                    if heavy_n < 2:
+                        heavy.append((start + duration, kind))
+                        placed = True
+                        stamp = start
+                        break
+                    nxt = min(item[0] for item in heavy)
+                    if nxt > arrival + 1.5:
+                        break
+                    start = nxt
+            elif mode == "O4" and kind != "consult":
+                heavy_jobs = sum(1 for _end, slot in heavy if slot != "consult")
+                if heavy_jobs < 1 and heavy_n < 2:
+                    heavy.append((arrival + duration, kind))
+                    placed = True
+            else:
+                if heavy_n < 2:
+                    heavy.append((arrival + duration, kind))
+                    placed = True
+            if placed:
+                accepted[kind] += 1
+            else:
+                rejected[kind] += 1
+            running, lights = occupy(stamp)
+            peak = max(peak, running + lights)
+
+        def rate(kind: str) -> str:
+            total = accepted[kind] + rejected[kind]
+            return "0" if total == 0 else f"{rejected[kind] / total:.2f}"
+
+        return {
+            "scheme": mode,
+            "consult_429": rate("consult"),
+            "report_429": rate("report"),
+            "score_429": rate("score"),
+            "peak": peak,
+            "touches_bug_717": mode == "O2",
+        }
+
+    return [run(mode) for mode in ("O1", "O2", "O3", "O4")]
+
+
+def resolve_record_path(explicit: str | None) -> Path:
+    """Write the A/B JSON only where asked. The committed note stays untouched."""
+    if explicit:
+        return Path(explicit)
+    directory = Path(tempfile.mkdtemp(prefix="api-chart-subprocess-ab-"))
+    return directory / "api-chart-subprocess-gate-ab-20261013.json"
+
+
+def main(argv: list[str] | None = None) -> None:
+    parser = argparse.ArgumentParser(description="A/B for issue #64 chart subprocess gate")
+    parser.add_argument(
+        "--out",
+        default=None,
+        help="Write the JSON record to this path. Omit to use a fresh temp directory, not the committed docs/testing file.",
+    )
+    args = parser.parse_args(argv)
+    counter = _Counter()
+    counter.install()
+    charts = _measure_charts(counter)
+    forced = _measure_report(counter, spawn=True)
+    native = _measure_report(counter, spawn=False)
+    report_match = _canon(forced["payload"]) == _canon(native["payload"])
+    schemes = _simulate()
+    out = {
+        "charts": charts,
+        "report_forced_subprocess": {key: value for key, value in forced.items() if key != "payload"},
+        "report_short_circuit": {key: value for key, value in native.items() if key != "payload"},
+        "report_bytes_match": report_match,
+        "schemes": schemes,
+        "jhora_installed": False,
+    }
+    text = json.dumps(out, ensure_ascii=False, indent=2)
+    dest = resolve_record_path(args.out)
+    dest.parent.mkdir(parents=True, exist_ok=True)
+    dest.write_text(text + "\n", encoding="utf-8")
+    print(dest, file=sys.stderr)
+    print(text)
+
+
+if __name__ == "__main__":
+    main()
diff --git a/scripts/tajika.py b/scripts/tajika.py
index 906b3495..4e6656c8 100644
--- a/scripts/tajika.py
+++ b/scripts/tajika.py
@@ -1501,7 +1501,28 @@ print(json.dumps({str(k): float(v) for k, v in dict(scores).items()}))
         return None
 
 
+_external_pyjhora_python_cached = False
+_external_pyjhora_python_value: Optional[str] = None
+
+
+def reset_external_pyjhora_python_cache() -> None:
+    """Tests and a fresh process both start with no remembered interpreter."""
+    global _external_pyjhora_python_cached, _external_pyjhora_python_value
+    _external_pyjhora_python_cached = False
+    _external_pyjhora_python_value = None
+
+
 def _resolve_external_pyjhora_python() -> Optional[str]:
+    """Remember the lookup, including "not installed", for this process."""
+    global _external_pyjhora_python_cached, _external_pyjhora_python_value
+    if _external_pyjhora_python_cached:
+        return _external_pyjhora_python_value
+    _external_pyjhora_python_value = _lookup_external_pyjhora_python()
+    _external_pyjhora_python_cached = True
+    return _external_pyjhora_python_value
+
+
+def _lookup_external_pyjhora_python() -> Optional[str]:
     candidates = [
         os.environ.get('JYOTISH_PYJHORA_PYTHON'),
         shutil.which('python3.14'),
diff --git a/tests/fixtures/api-chart-subprocess-gate/README.txt b/tests/fixtures/api-chart-subprocess-gate/README.txt
new file mode 100644
index 00000000..e5178a32
--- /dev/null
+++ b/tests/fixtures/api-chart-subprocess-gate/README.txt
@@ -0,0 +1,6 @@
+Generated 2026-10-15 by calling the isolated workers with PyJHora absent
+(importlib.util.find_spec("jhora") is None on the Windows venv):
+
+  .venv\Scripts\python -c "import json; from scripts.annual_pyjhora_replay import _isolated_tajaka_worker, _isolated_annual_worker; req={'birth_julian_day':2447893.0,'place':{'name':'Delhi','latitude':28.6139,'longitude':77.209,'timezone':5.5,'elevation':0},'age':30,'settings':{'ayanamsa':'LAHIRI','node_mode':'mean'}}; print(json.dumps(_isolated_tajaka_worker(req), sort_keys=True)); print(json.dumps(_isolated_annual_worker(req), sort_keys=True, default=str))"
+
+The parent adds request_hash after this body. The short-circuit must match this body.
diff --git a/tests/fixtures/api-chart-subprocess-gate/annual-no-jhora.json b/tests/fixtures/api-chart-subprocess-gate/annual-no-jhora.json
new file mode 100644
index 00000000..df78a16f
--- /dev/null
+++ b/tests/fixtures/api-chart-subprocess-gate/annual-no-jhora.json
@@ -0,0 +1,41 @@
+{
+  "annual_chart_snapshot": {
+    "reason": "pyjhora_annual_adapter_setup_failed:ModuleNotFoundError",
+    "status": "blocked"
+  },
+  "annual_chart_speed_snapshot": {
+    "reason": "pyjhora_annual_adapter_setup_failed:ModuleNotFoundError",
+    "status": "blocked"
+  },
+  "effective_ayanamsa": null,
+  "evidence_scope": "pyjhora_behavior_only",
+  "mudda_dasha": {
+    "reason": "pyjhora_annual_adapter_setup_failed:ModuleNotFoundError",
+    "status": "blocked"
+  },
+  "node_mode": {
+    "requested": "mean",
+    "status": "unsupported_by_replay_adapter"
+  },
+  "parity_status": "not_multiengine_parity",
+  "patyayini_dasha": {
+    "reason": "pyjhora_annual_adapter_setup_failed:ModuleNotFoundError",
+    "status": "blocked"
+  },
+  "sahams": {
+    "reason": "pyjhora_annual_adapter_setup_failed:ModuleNotFoundError",
+    "status": "blocked"
+  },
+  "settings": {
+    "ayanamsa": "LAHIRI",
+    "node_mode": "mean"
+  },
+  "solar_return_boundary": {
+    "reason": "pyjhora_annual_adapter_setup_failed:ModuleNotFoundError",
+    "status": "blocked"
+  },
+  "tajika_yogas": {
+    "reason": "pyjhora_annual_adapter_setup_failed:ModuleNotFoundError",
+    "status": "blocked"
+  }
+}
diff --git a/tests/fixtures/api-chart-subprocess-gate/tajaka-no-jhora.json b/tests/fixtures/api-chart-subprocess-gate/tajaka-no-jhora.json
new file mode 100644
index 00000000..2c1534ae
--- /dev/null
+++ b/tests/fixtures/api-chart-subprocess-gate/tajaka-no-jhora.json
@@ -0,0 +1,12 @@
+{
+  "node_mode": {
+    "requested": "mean",
+    "status": "unsupported_by_replay_adapter"
+  },
+  "reason": "pyjhora_ayanamsa_configuration_failed:ModuleNotFoundError",
+  "settings": {
+    "ayanamsa": "LAHIRI",
+    "node_mode": "mean"
+  },
+  "status": "blocked"
+}
diff --git a/tests/test_api_chart_subprocess_gate.py b/tests/test_api_chart_subprocess_gate.py
new file mode 100644
index 00000000..e8a97821
--- /dev/null
+++ b/tests/test_api_chart_subprocess_gate.py
@@ -0,0 +1,197 @@
+"""Issue #64: skip useless report subprocesses and observe the heavy-compute gate."""
+
+from __future__ import annotations
+
+import json
+import logging
+from pathlib import Path
+
+import pytest
+
+from scripts import api_heavy_compute_gate as gate
+from scripts.annual_pyjhora_replay import (
+    _run_isolated_annual_replay,
+    _run_isolated_tajaka_year_lord,
+    reset_jhora_import_cache,
+)
+from scripts.jyotish_engine import (
+    _oracle_progress_snapshot,
+    reset_oracle_progress_snapshot_cache,
+)
+from scripts.tajika import (
+    _resolve_external_pyjhora_python,
+    reset_external_pyjhora_python_cache,
+)
+
+ROOT = Path(__file__).resolve().parents[1]
+FIXTURES = ROOT / "tests" / "fixtures" / "api-chart-subprocess-gate"
+
+
+class _Place:
+    name = "Delhi"
+    latitude = 28.6139
+    longitude = 77.209
+    timezone = 5.5
+    elevation = 0
+
+
+_PROFILE = {"ayanamsa": "LAHIRI", "node_mode": "mean"}
+
+
+def _without_hash(payload: dict) -> dict:
+    body = dict(payload)
+    body.pop("request_hash", None)
+    return body
+
+
+def test_no_jhora_tajaka_skips_subprocess_and_matches_golden(monkeypatch) -> None:
+    reset_jhora_import_cache()
+    monkeypatch.setattr("scripts.annual_pyjhora_replay._jhora_importable", lambda: False)
+
+    def _forbid_subprocess(*_args, **_kwargs):
+        raise AssertionError("subprocess.run must not run when jhora cannot be imported")
+
+    monkeypatch.setattr("scripts.annual_pyjhora_replay.subprocess.run", _forbid_subprocess)
+    result = _run_isolated_tajaka_year_lord(_PROFILE, 2447893.0, _Place(), 30)
+    golden = json.loads((FIXTURES / "tajaka-no-jhora.json").read_text(encoding="utf-8"))
+    assert result["request_hash"]
+    assert _without_hash(result) == golden
+
+
+def test_no_jhora_annual_skips_subprocess_and_matches_golden(monkeypatch) -> None:
+    reset_jhora_import_cache()
+    monkeypatch.setattr("scripts.annual_pyjhora_replay._jhora_importable", lambda: False)
+
+    def _forbid_subprocess(*_args, **_kwargs):
+        raise AssertionError("subprocess.run must not run when jhora cannot be imported")
+
+    monkeypatch.setattr("scripts.annual_pyjhora_replay.subprocess.run", _forbid_subprocess)
+    result = _run_isolated_annual_replay(_PROFILE, 2447893.0, _Place(), 30)
+    golden = json.loads((FIXTURES / "annual-no-jhora.json").read_text(encoding="utf-8"))
+    assert result["request_hash"]
+    assert _without_hash(result) == golden
+
+
+def test_jhora_available_still_uses_subprocess(monkeypatch) -> None:
+    reset_jhora_import_cache()
+    monkeypatch.setattr("scripts.annual_pyjhora_replay._jhora_importable", lambda: True)
+    calls = {"n": 0}
+
+    class _Completed:
+        returncode = 0
+        stdout = json.dumps({"status": "partial_verified", "settings": {}})
+
+    def _fake_run(*_args, **_kwargs):
+        calls["n"] += 1
+        return _Completed()
+
+    monkeypatch.setattr("scripts.annual_pyjhora_replay.subprocess.run", _fake_run)
+    tajaka = _run_isolated_tajaka_year_lord(_PROFILE, 2447893.0, _Place(), 30)
+    annual = _run_isolated_annual_replay(_PROFILE, 2447893.0, _Place(), 30)
+    assert calls["n"] == 2
+    assert tajaka["request_hash"]
+    assert annual["request_hash"]
+
+
+def test_pyjhora_interpreter_lookup_is_cached(monkeypatch) -> None:
+    reset_external_pyjhora_python_cache()
+    calls = {"n": 0}
+
+    def _missing():
+        calls["n"] += 1
+        return None
+
+    monkeypatch.setattr("scripts.tajika._lookup_external_pyjhora_python", _missing)
+    assert _resolve_external_pyjhora_python() is None
+    assert _resolve_external_pyjhora_python() is None
+    assert calls["n"] == 1
+
+
+def test_oracle_progress_snapshot_reuses_until_the_file_changes(monkeypatch) -> None:
+    reset_oracle_progress_snapshot_cache()
+    calls = {"n": 0}
+    real_run = __import__("subprocess").run
+
+    def _counting_run(*args, **kwargs):
+        calls["n"] += 1
+        return real_run(*args, **kwargs)
+
+    monkeypatch.setattr("subprocess.run", _counting_run)
+    first = _oracle_progress_snapshot()
+    after_first = calls["n"]
+    assert after_first >= 1
+    second = _oracle_progress_snapshot()
+    assert calls["n"] == after_first
+    assert second == first
+
+    oracle = ROOT / "references" / "oracle" / "dasha_shadbala_oracle_cases.json"
+    stat = oracle.stat()
+    later = stat.st_mtime + 5
+    try:
+        oracle.touch()
+        import os
+        os.utime(oracle, (later, later))
+        third = _oracle_progress_snapshot()
+        assert calls["n"] > after_first
+        assert third["scope"] == first["scope"]
+    finally:
+        os_utime = __import__("os").utime
+        os_utime(oracle, (stat.st_atime, stat.st_mtime))
+        reset_oracle_progress_snapshot_cache()
+
+
+def test_chart_overview_gate_is_off_by_default(monkeypatch) -> None:
+    monkeypatch.delenv(gate.ENV_GATE_CHART_OVERVIEW, raising=False)
+    gate.reset_heavy_compute_gate(limit=1)
+    held = gate.acquire_heavy_compute_slot("/api/high_rigor_workflow")
+    assert held is not None
+    assert gate.acquire_heavy_compute_slot_for_request("/api/chart", {"year": 1990}) is None
+    gate.release_heavy_compute_slot(held)
+    assert gate.heavy_compute_status()["in_use"] == 0
+    gate.reset_heavy_compute_gate()
+
+
+def test_chart_overview_gate_only_sync_unskipped_requests(monkeypatch) -> None:
+    monkeypatch.setenv(gate.ENV_GATE_CHART_OVERVIEW, "1")
+    gate.reset_heavy_compute_gate(limit=1)
+    body = {"year": 1990, "lat": 28.6, "lon": 77.2}
+    slot = gate.acquire_heavy_compute_slot_for_request("/api/chart", body)
+    assert slot is not None
+    assert gate.heavy_compute_status() == {"limit": 1, "in_use": 1}
+    with pytest.raises(gate.HeavyComputeBusy):
+        gate.acquire_heavy_compute_slot_for_request("/api/chart", {"year": 1991})
+    gate.release_heavy_compute_slot(slot)
+    assert gate.heavy_compute_status()["in_use"] == 0
+    assert gate.acquire_heavy_compute_slot_for_request(
+        "/api/chart", {"skip_vedastro_main_entry_overview": True}
+    ) is None
+    assert gate.acquire_heavy_compute_slot_for_request("/api/chart", {"async": True}) is None
+    assert gate.acquire_heavy_compute_slot_for_request("/api/chart", {"enqueue": 1}) is None
+    assert "/api/chart" not in gate.HEAVY_COMPUTE_PATHS
+    gate.reset_heavy_compute_gate()
+
+
+def test_busy_log_names_holders_and_not_the_body(caplog) -> None:
+    gate.reset_heavy_compute_gate(limit=1)
+    held = gate.acquire_heavy_compute_slot("/api/professional_report_reference")
+    assert held is not None
+    secret_body = {"birth_date": "1990-01-01", "name": "must-not-appear"}
+    with caplog.at_level(logging.WARNING, logger="jyotish.heavy_compute"):
+        with pytest.raises(gate.HeavyComputeBusy):
+            gate.acquire_heavy_compute_slot_for_request("/api/professional_report_reference", secret_body)
+    text = caplog.text
+    assert "/api/professional_report_reference" in text
+    assert "1990-01-01" not in text
+    assert "must-not-appear" not in text
+    gate.release_heavy_compute_slot(held)
+    assert gate.heavy_compute_status()["in_use"] == 0
+    gate.reset_heavy_compute_gate()
+
+
+def test_warm_consultation_still_skips_the_slot() -> None:
+    gate.reset_heavy_compute_gate(limit=1)
+    assert gate.acquire_heavy_compute_slot_for_request(
+        "/api/consultation_workflow", {"consult_card_warm_only": True}
+    ) is None
+    assert gate.heavy_compute_status()["in_use"] == 0
+    gate.reset_heavy_compute_gate()
diff --git a/tests/test_supabase_user_data_contract.py b/tests/test_supabase_user_data_contract.py
index e4300dbd..63633b2b 100644
--- a/tests/test_supabase_user_data_contract.py
+++ b/tests/test_supabase_user_data_contract.py
@@ -298,8 +298,11 @@ def test_synastry_route_orchestrates_python_chart_and_ashtakoot() -> None:
     route = SYNASTRY_ROUTE.read_text(encoding="utf-8")
     for token in (
         'const apiBase = process.env.JYOTISH_API_BASE ?? "http://127.0.0.1:5200"',
-        'postPython("/api/chart", selfPayload)',
-        'postPython("/api/chart", partnerPayload)',
+        # 原值: postPython("/api/chart", selfPayload) 与 postPython("/api/chart", partnerPayload)
+        # 新值: 两次调用都展开对应 payload，并带 skip_vedastro_main_entry_overview: true
+        # 原因: 合盘改为同时排本人和对方两张盘，并关掉用不上的外部概览。断言仍锁住这两次 /api/chart，同时锁住这个开关，不放宽。
+        'postPython("/api/chart", { ...selfPayload, skip_vedastro_main_entry_overview: true })',
+        'postPython("/api/chart", { ...partnerPayload, skip_vedastro_main_entry_overview: true })',
         'postPython("/api/varga_full"',
         'postPython("/api/synastry"',
         "moonLongitude(selfChart)",