feat(pmflow-os): connect execution capability gate

docs/PMFLOW_PRODUCT_DELIVERY_STANDARD.md

@@ -765,113 +765,85 @@ commit gate 判定：
 
 当前已完成：
 
-**P8：Review Decision Loop 剩余消费面收口。**
+**main-flow-d4-execution-runtime-capability：Execution Capability Graph 接入开发执行页。**
 
 原因：
 
-- Main Flow 已把 `reviewerRouting` 从“只读建议”推进到 `review request -> reviewer decision -> approval audit`，并复用 `PMFlowApproval` 形成审计记录。
+- `/execution-bridge` 已从字段调试页推进到产品化开发执行工作面，并进一步消费 `execution-capabilities`，把 `AI 直接执行 / AI 执行 + 用户验收 / AI 执行 + 技术审查 / 需要人类先补判断 / 当前不可执行` 变成首屏执行模式判定。
-- Space 已把待审/已请求/已决策事项纳入 `/space` 的个人下一步队列，Account 已补 review decision / high-risk approval 的页面内提醒偏好合同。
+- PMFlow OS store 已把 `GET /pmflow-os/projects/{project_id}/execution-capabilities` 纳入共享状态，执行交接除正式 Task gate 外，也会在 reviewer 缺失时阻断 handoff。
-- P8 集成仍不新增战略页面，不改 router/nav，不接真实通知中心；继续使用既有 `/change-review / /team-layer / /space / /me` 表面。
+- `/execution-bridge` 已展示“本次执行结果”汇总，把运行端、代码证据、测试证据、阻塞与下一步收敛成验收前判断，不把缺证据状态表达成发布成功。
 
-### 7.1 P8 并行建议
+### 7.1 下一步推荐 Packet：`main-flow-d5-execution-acceptance-loop`
 
-P8 可以三条 lane 并行，但共享高冲突文件必须由 Main Flow 或 Shell / Integration 串行处理。
+状态：建议下一步。
-
-#### Main Flow Packet：`main-flow-p8-review-decision-loop`
-
-状态：已完成。
 
 目标：
 
-- 把 reviewerRouting 从“只读建议”推进到“审查决策闭环”。
+- 把 `execution_result_protocol_v1` 回写后的结果接到“用户验收 / 技术审查 / 发布门禁”状态流。
-- 支持对待审 ChangeSet 请求审查、记录 reviewer decision、形成 `PMFlowApproval` 审计记录。
+- 区分 `待回写 / 执行中 / 执行受阻 / 待用户验收 / 待技术审查 / 已验收`，并让 Space next action 在验收后能收敛。
-- 让 `ChangeReviewPage / TeamLayerPage` 展示审查状态、审查意见、高风险 approval gate 和下一步动作。
+- 技术审查模式必须保留 reviewer gate；agent / IDE / Local Bridge 回写测试通过，不等于自动完成。
-- reviewer decision 只影响审查/approval 状态，不自动 commit ChangeSet。
+- 通过验收后再让 Release/Growth 门禁读取稳定状态，缺 PR、测试或阻塞证据时继续 fail-closed。
-- high-risk ChangeItem 必须继续遵守 C2 approval gate。
 
 建议写范围：
 
 - `backend/app/api/pmflow_os_api/`
-- `backend/app/services/pmflow_change_review_service.py`
 - `backend/app/services/pmflow_os_projection_service.py`
 - `backend/app/schemas/pmflow_os.py`
-- `backend/tests/test_pmflow_change_review_api.py`
 - `backend/tests/test_pmflow_os_projection_api.py`
 - `frontend/src/api/pmflowOs.ts`
 - `frontend/src/features/pmflow-os/store.tsx`
 - `frontend/src/features/pmflow-os/store.api.test.tsx`
-- `frontend/src/features/pmflow-os/types.ts`
+- `frontend/src/features/pmflow-os/pages/ExecutionBridgePage.tsx`
-- `frontend/src/features/pmflow-os/pages/ChangeReviewPage.tsx`
+- `frontend/src/features/pmflow-os/pages/ReleaseGrowthPage.tsx`
-- `frontend/src/features/pmflow-os/pages/TeamLayerPage.tsx`
+- 必要文档
-- 相关 Main Flow 文档
 
 禁止：
 
-- 不新增 migration，优先复用 `PMFlowApproval`。
+- 不执行本机命令、不推送、不发布。
-- 不新增 route/nav。
+- 不自动创建正式任务。
-- 不改 Space / Account。
+- 不绕过 Project Memory commit 后才能执行的边界。
-- 不自动 commit ChangeSet。
+- 不把测试通过自动等同于用户验收或技术审查通过。
-- 不接真实邮件、桌面推送或系统通知中心。
-
-#### Space Packet：`space-p7-review-attention-queue`
-
-状态：已完成。
-
-目标：
-
-- 把 Team Review 的待审事项接入 `/space` 的“我的下一步”队列。
-- 只读消费 `truth-state.reviewerRouting`；如果 Main Flow P8 已完成，再只读消费 `reviewDecision / approvalState`。
-- 展示“我需要审什么 / 哪些审查卡住了项目”，所有动作仍只跳主流程。
-
-建议写范围：
-
-- `frontend/src/api/space.ts`
-- `frontend/src/features/personal-space/`
-- `frontend/src/features/personal-space/PersonalSpacePage.test.tsx`
-- Space 相关文档
-
-禁止：
-
-- 不写 ChangeSet、`PMFlowApproval`、ProgressEvent、ReleaseValidationRun、FeedbackIngest。
-- 不改 PMFlow OS、Account、router 或 bottom nav。
-- 不虚构 reviewer 或待审项。
-
-#### Account Settings Packet：`account-p8-review-decision-preference`
-
-状态：已完成。
-
-目标：
-
-- 扩展 Account Settings 的提醒偏好合同，增加 review decision / reviewer request / high-risk approval 相关偏好。
-- 仍然只控制页面内提醒强度，不接真实通知中心。
-- 偏好关闭只能静默强调，不得隐藏真实 reviewerRouting、approval、evidence 或 audit state。
-
-建议写范围：
-
-- `frontend/src/api/accountSettings.ts`
-- `frontend/src/features/account-settings/AccountSettingsPanel.tsx`
-- `frontend/src/features/account-settings/accountSettings.test.ts`
-- Account 相关文档
-
-禁止：
-
-- 不改 PMFlow OS、Space、router 或 backend schema。
-- 不把偏好当业务权限。
-
-### 7.2 P8 Integration Owner 收口
-
-状态：已完成。
-
-目标：
-
-- 检查 Main Flow / Space / Account 三个 Packet 的 `git diff --name-only` 是否越界。
-- 统一处理共享文档、API 索引和状态台账冲突。
-- 确认 P8 仍不新增战略页面，仍使用既有 `/change-review / /team-layer / /space / /me` 表面。
-- 跑 Main Flow / Space / Account focused tests、全量 typecheck/build，并在后端变更后重启 backend 确认 healthy。
-- 更新 `## 8. 进度台账`、`CODE_REFERENCE_INDEX.md`、`ARCHITECTURE_INVENTORY.md` 与必要的 `CLAUDE.md`。
 
 ## 8. 进度台账
 
+### 2026-05-05 / main-flow-d4-execution-runtime-capability / implemented
+
+完成内容：
+
+- PMFlow OS store 新增真实 `execution-capabilities` 查询状态，`refetchApiState` 会同步刷新 truth-state、execution-summary 与 execution-capabilities。
+- `createExecutionHandoff` 在正式任务 gate 之外新增 reviewer/capability gate：当技术审查责任缺失时阻止真实 handoff，提示先回团队层或待确认中心补齐审查责任。
+- `/execution-bridge` 首屏新增“执行模式判定”，消费 Execution Capability Graph 展示 `AI 直接执行 / AI 执行 + 用户验收 / AI 执行 + 技术审查 / 需要人类先补判断 / 当前不可执行`。
+- `/execution-bridge` 新增“本次执行结果”汇总，把运行端、代码证据、测试证据、阻塞与下一步收敛为验收前判断；缺证据时保持待回写，不进入发布成功表达。
+- 执行交接主按钮改为能力 gate：候选/草稿任务、待确认范围、缺 reviewer 会阻塞；缺 repo 时只允许生成交接草案；缺 Prompt Package 时提示生成上下文包。
+- AI 副驾从泛审查文案升级为执行裁决建议，只基于 execution-summary、execution-capabilities、progress event 与 release gate，不调用新 LLM，不编造 PR/测试/代码质量。
+
+涉及文件：
+
+- `frontend/src/features/pmflow-os/store.tsx`
+- `frontend/src/features/pmflow-os/pages/ExecutionBridgePage.tsx`
+- `frontend/src/features/pmflow-os/store.api.test.tsx`
+- `docs/PMFLOW_PRODUCT_DELIVERY_STANDARD.md`
+- `docs/architecture/CODE_REFERENCE_INDEX.md`
+- `docs/architecture/ARCHITECTURE_INVENTORY.md`
+
+验证：
+
+- `cd frontend && npx tsc --noEmit --pretty false --incremental false` 通过。
+- `cd frontend && npm run test -- src/features/pmflow-os/store.api.test.tsx -t "execution capability|technical review|execution bridge"` 通过：2 passed，32 skipped。
+- `cd frontend && npm run test -- src/features/pmflow-os` 通过：35 passed；仍有既有 React Router future flag warning。
+- `cd frontend && npm run build` 通过。
+- Playwright / Chromium 打开 `http://localhost:5176/execution-bridge?demo=1`，桌面 1440x1100 与移动 390x844 均能渲染“执行模式判定 / 本次执行结果 / 执行交接”模块；截图输出到 `output/playwright/execution-bridge-d4-desktop.png` 与 `output/playwright/execution-bridge-d4-mobile.png`。
+- 本轮只修改前端和文档，未修改后端代码、Schema 或 migration，因此未执行后端编译、数据库迁移或热重启。
+
+未完成或风险：
+
+- 当前未接真实 Deep Agents runtime，也不执行本机命令；Execution Bridge 仍只做上下文包、handoff 合同与结果回写展示。
+- 浏览器实测中本地无后端认证导致既有 `GET /api/ai-readiness/project-launch` 401，这属于项目壳 AI readiness 本地环境问题，不影响本页模块渲染结论。
+
+下一步建议：
+
+- 进入 `main-flow-d5-execution-acceptance-loop`：把执行结果回写接到用户验收 / 技术审查 / 发布门禁状态流，避免测试通过后仍停留在“待回写”或被误认为自动完成。
+
 ### 2026-05-05 / main-flow-d3-execution-bridge-productization / implemented
 
 完成内容：

docs/architecture/ARCHITECTURE_INVENTORY.md

@@ -213,6 +213,7 @@
 - `frontend/src/features/pmflow-os/store.tsx` 当前是 PMFlow OS 共享 truth store：`?projectId=<id>` 时通过 React Query 调用 `frontend/src/api/pmflowOs.ts` 读取真实 `GET /pmflow-os/projects/{project_id}/truth-state` 与 `GET /pmflow-os/projects/{project_id}/execution-summary`，真实模式下项目内 `sendMessage` 会写 thread message 并触发后端真实 AI extraction；显式 `proposed_items` extraction、ChangeItem `accept/edit/defer`、ChangeSet commit、review decision、Execution Handoff create/consume、execution progress writeback、release validation run create、release feedback ingest、单条 iteration recommendation -> ChangeSet draft 与同组 iteration recommendation 批量 -> ChangeSet draft 均调用 `/pmflow-os` 后端并 refetch；只有显式 `?demo=1` 或 `?dataSource=demo` 时才启用本地 mock 演示状态机。真实读取/动作失败会进入错误/空态，不静默 fallback 成 mock 成功；AI extraction 失败显示“结构化失败/待补充”，不创建模板 ChangeSet。
 - `frontend/src/features/pmflow-os/store.tsx` 仍保留演示用项目内 AI 结构化闭环状态机：`command_idle / ai_extracting / extraction_ready / review_open / writeback_success`。`frontend/src/features/pmflow-os/pages/StructuredExtractionPage.tsx` 是本轮结构化提取页，也是结构化读取检查层：从指挥台进入时主动作进入待确认或继续补充；从 `ChangeReviewPage.tsx` 进入时通过 `origin=change_review / returnTo=change_review / changeSetId / itemId` 保留审查上下文，主动作返回待确认队列，补充证据才回指挥台；已 committed 的轮次显示为结构化写入快照并转向项目记忆或新变更链路。`ChangeReviewPage.tsx` 是决策与待确认中心，真实模式下顶部识别统计、待确认列表与右侧 AI 副驾评估都只消费后端 `changeSets / changeItems / evidence`，没有真实待审项时显示空态，禁止用 demo 审查项补位；确认写入 Project Memory 前，`pending_review / accepted / edited / deferred` ChangeItem 均可继续 accept/edit/defer，只有 `committed / skipped` 或关闭轮次不可再操作；`MemoryWritebackResultPage.tsx` 是项目记忆写入结果页，`WorkGraphPage.tsx` 是项目工作台关系总览与对象投影视图，`ExecutionBridgePage.tsx` 是执行桥上下文包、交接按钮、运行端领取确认、PR/测试与回写状态页；`ReleaseGrowthPage.tsx` 真实模式读取 `execution-summary.releaseGate`、`latestValidationRun` 与 `releaseSignals`，从 branch、commit、PR、test、failed evidence 展示 `可发布 / 阻塞 / 需要人工确认 / 缺少证据` 判断、阻塞原因、下一步动作、发布验证记录和发布后 feedback/revenue/ROI/compliance/recommendation 信号，缺少发布后信号时显示 empty/warning，不执行发布命令、不展示 mock 发布结果；页面可展示 recommendation quality grouping，把真实 recommendation 单条或同组批量生成下一轮待确认 ChangeSet draft，生成后仍进入 Change Review accept/edit/defer，不自动 commit、不写 Project Memory、不创建正式任务；`ChangeReviewPage.tsx` 会展示 `iteration_recommendation_group` evidence 的来源组、feedback 数、recommendation 数、风险、影响和审查优先级；同时读取 Account notification preference helper，用 `releaseGate` 偏好控制发布门禁强调提醒，用 `failureAlert` 偏好控制失败强调提醒，偏好关闭时只静默提醒表达强度，不隐藏 release evidence、阻塞原因、审计状态、validation run 或 releaseSignals；静态发布/增长原型只保留在显式 demo 模式。A3 已接入后端 truth-state，C1 已接入真实 message/extraction/review 状态机，C2 已接入真实 ChangeSet commit，Phase D 已接入真实 execution-summary/dev-sync，D.2 已接入 `POST /execution-handoffs` 创建 PromptPackage + handoff_created progress event，P1 已接入 `POST /execution-handoffs/{prompt_package_id}/consume` 创建 handoff_consumed progress event 并把 execution status 投影为 coding，Release Growth P1 已接入 execution evidence gate，Release Validation P1 已接入 `POST /release-validation-runs` 创建 `ReleaseValidationRun` 快照，Release Feedback P1 已接入 `POST /release-feedback-ingests` 写入 `FeedbackIngest` 并生成 `IterationRecommendation`，Iteration Loop P1 已接入 `POST /iteration-recommendations/{recommendation_id}/change-set-drafts` 把 recommendation 回流到 Change Review，Iteration Loop P2 已接入 `POST /iteration-recommendation-groups/{group_key}/change-set-drafts` 并在 `releaseSignals.recommendationGroups` 暴露分组/去重/风险/影响合同；真实 AI extraction 已接入独立 `change_review` protected scene、`pmflow_change_review_extraction_v1` 运行合同、`PROMPT_PMFLOW_CHANGE_REVIEW_EXTRACTION` Prompt 资产与后端 `llm_service.generate_structured_object`，成功时生成 open ChangeSet、pending_review ChangeItems 与 thread message evidence，失败时 fail-closed，不把 mock 当真实失败 fallback，也不复用 Project Launch draft prompt/function。`ProjectMemoryPage.tsx` 在真实模式下展示后端 `truth_nodes + semantic_facts` 的 `projectMemory` 投影，accepted/edited ChangeItem 只有 commit 成功后才进入 Project Memory。`frontend/src/features/pmflow-os/components/ProjectStructuredLoop.tsx` 提供 `StatusLabel / EvidenceSnippet / ObjectChangeCard / ProjectConversationPanel / StructuredExtractionPanel / ProjectAICopilotPanel` 等闭环组件，并复用契约化 `InlineNotice` 表达“未确认不得写入”的局部提示。
 - Execution Result Protocol P1 将 Execution Bridge 的执行结果回写收敛为 `execution_result_protocol_v1`：`backend/app/schemas/pmflow_os.py` 接收 `execution_status` 与 `blocker_summary`，`backend/app/services/pmflow_os_projection_service.py` 写入 `ProgressEvent.payload.protocol_version` 与 `execution_result` envelope，同时保留 `branch_name / commit_sha / pr_number / test_summary / workflow_state` flat 字段兼容现有投影；`frontend/src/api/pmflowOs.ts` 和 `ExecutionBridgePage.tsx` 支持手动记录 status/blocker。该链路只记录 Codex/IDE/Local Bridge/GitHub agent 已产生的结果，不执行本机命令、不推送、不发布、不自动创建正式任务；`execution-summary`、`releaseGate` 与 Team Feed 均从 `ProgressEvent` 只读消费。
+- Main Flow D4 已把 Execution Capability Graph 接入 `/execution-bridge`：`frontend/src/features/pmflow-os/store.tsx` 现在读取并刷新真实 `GET /pmflow-os/projects/{project_id}/execution-capabilities`，`ExecutionBridgePage.tsx` 首屏展示执行模式判定、本次执行结果、执行链路、上下文包、回写视图和 AI 执行裁决建议；handoff 主按钮受正式 Task、待确认范围、repo/prompt package 与 reviewer gate 约束，技术审查责任缺失时 store 会阻止真实 handoff，缺 repo 时只允许表达交接草案。该层仍只做能力判定、交接合同和结果回写展示，不执行本机命令、不触发真实 agent、不自动验收、不自动发布。
 - Project Knowledge Hub 读模型由 `GET /pmflow-os/projects/{project_id}/project-memory/overview`、`items`、`items/{item_id}`、`assets`、`context-packs`、`writebacks` 承载，后端在 `backend/app/services/pmflow_os_projection_service.py` 从 `truth_nodes / tasks / semantic_facts / pmflow_evidence_links / pmflow_thread_messages / pmflow_change_sets / pmflow_change_items` 聚合首页雷达、全部记忆、项目资产、上下文包和写入版本记录。前端对应 `ProjectMemoryPage.tsx`、`ProjectMemoryCollectionPages.tsx` 与 `projectMemoryUtils.ts`：列表时间使用短格式，抽屉显示完整时间和北京时间；首页卡片与二级列表单条点击打开右侧详情抽屉，展示来源证据、影响对象和 AI 执行动作跳转；全部记忆支持类型、阶段、来源、状态、搜索筛选。该读模型只读，资产可自动索引，进入项目真相层仍必须经过 Change Review 或可信规则确认。
 - `frontend/src/features/account-settings/AccountSettingsPanel.tsx` 是 `/me` 设置页实现：桌面端为个人资料摘要 + 分区设置面板，移动端为分组列表 + `MobileBottomSheet` 编辑；当前覆盖个人资料、默认语言、AI 回答深度、证据展示偏好、待确认、执行回写、发布门禁和失败提醒通知偏好，并在通知偏好区展示页面内提醒消费说明、提醒强度和证据展示配置。`frontend/src/api/accountSettings.ts` 是前端合同层；有 token 时优先调用 `backend/app/api/account_settings_api/`，失败时回落浏览器本机暂存；P2 新增 `shouldShowAccountNotification` 与 `getAccountNotificationConsumption`，将四类通知偏好映射为本地 UI 的待确认、Execution Bridge、Release/Growth 和失败提醒显示/静默状态；P3 新增 `getReleaseGateNotificationPreference(settings, event)`，把 Release/Growth 的 `release_gate_* / release_validation_*` 事件映射到 `releaseGate` 偏好，把 `failed_event` 映射到 `failureAlert` 偏好；P5/P6 的 `getPostReleaseNotificationPreference(settings, event)` 把发布后 `feedback_received / revenue_signal / compliance_blocked` 映射到 `releaseGate` 偏好，把 `iteration_recommendation / next_change_review` 映射到 `pendingConfirmation` 偏好和 `change_review` surface，并返回 `spaceReadable / evidencePolicy / reminderPolicy` 供 Main Flow 与 Space 消费；P7 将通知偏好扩展为 `reminderIntensity=silent|normal|prominent` 与 `evidenceDisplay=compact|expanded`，并新增 `getAccountReminderConsumption(settings, event)` 覆盖 `change_review_high_risk / iteration_draft_ready / release_blocked / execution_stalled`，输出 `notificationTransport=in_app_only / evidencePolicy=preserve_business_evidence / auditPolicy=preserve_audit_state`；P8 新增 `getReviewDecisionNotificationPreference(settings, event)` 覆盖 `review_requested / review_approved / review_changes_requested / review_rejected / high_risk_approval_required`，继续使用 `pendingConfirmation` 偏好和 `change_review` surface，并声明 reviewerRouting、approval、evidence、audit state 均必须保留。该合同只控制页面内提醒表达，不代表邮件、桌面推送或系统通知中心已接入；偏好关闭也不得隐藏业务证据、reviewerRouting、approval、recommendation、ChangeSet、执行状态或审计状态。`backend/app/schemas/account_settings.py` 定义账号偏好合同，`backend/app/services/account_settings_service.py` 当前复用既有 `integrations` 表按 `tenant_id + user_id` 保存 `account_settings_preferences` 配置包，`backend/app/api/router.py` 已挂载该 API；未改用户表、OAuth/JWT/auth core 或 PMFlow OS truth store。
 - `frontend/src/features/pmflow-os/store.tsx` 当前是 PMFlow OS 共享 truth store：`?projectId=<id>` 时通过 React Query 调用 `frontend/src/api/pmflowOs.ts` 读取真实 `GET /pmflow-os/projects/{project_id}/truth-state` 与 `GET /pmflow-os/projects/{project_id}/execution-summary`，真实模式下项目内 `sendMessage` 会写 thread message 并触发后端真实 AI extraction；显式 `proposed_items` extraction、ChangeItem `accept/edit/defer`、ChangeSet commit、review decision、AI feedback、Execution Handoff create/consume、execution progress writeback、release validation run create、release feedback ingest 与 iteration recommendation -> ChangeSet draft 均调用 `/pmflow-os` 后端并 refetch；AI feedback 写入 `pmflow_ai_feedback_events`，在 `truth-state.aiFeedback` 恢复当前用户选择，并由后端按 `prompt_key` 生成待复核 `JudgeRun / OptimizerCandidate` 质量优化候选，用户改回赞同时挂起此前候选，Prompt Asset quality summary 会聚合 PMFlow 反馈与优化候选；只有显式 `?demo=1` 或 `?dataSource=demo` 时才启用本地 mock 演示状态机。真实读取/动作失败会进入错误/空态，不静默 fallback 成 mock 成功；AI extraction 失败显示“结构化失败/待补充”，不创建模板 ChangeSet。

frontend/src/features/pmflow-os/pages/ExecutionBridgePage.tsx

@@ -17,6 +17,7 @@ import {
   Send,
   Server,
   Sparkles,
+  ShieldCheck,
   Terminal,
   TestTube2,
 } from "lucide-react";
@@ -41,6 +42,11 @@ import {
   PacketBlock,
 } from "@/features/pmflow-os/components/ExecutionBridgeCards";
 import { usePMFlowOsStore } from "@/features/pmflow-os/store";
+import type {
+  PMFlowExecutionCapabilityMode,
+  PMFlowExecutionCapabilitySnapshot,
+  PMFlowExecutionModeCapability,
+} from "@/features/pmflow-os/types";
 import { cn } from "@/lib/utils";
 
 const STATUS_LABEL: Record<string, string> = {
@@ -86,6 +92,7 @@ const TAB_ITEMS: Array<{ key: ExecutionTabKey; label: string }> = [
 
 type ExecutionTabKey = "changes" | "terminal" | "commits" | "pr" | "tests" | "writeback";
 type PipelineTone = "done" | "current" | "waiting" | "blocked";
+type CapabilityTone = "ready" | "partial" | "blocked" | "unknown";
 
 function isFormalDevTask(task?: PMFlowDevTask | null) {
   const status = String(task?.status || "").toLowerCase();
@@ -96,6 +103,122 @@ function isFormalDevTask(task?: PMFlowDevTask | null) {
   );
 }
 
+function primaryExecutionMode(capabilities?: PMFlowExecutionCapabilitySnapshot | null): PMFlowExecutionModeCapability | null {
+  if (!capabilities?.executionModes.length) return null;
+  const order: PMFlowExecutionCapabilityMode[] = [
+    "ai_with_technical_review",
+    "ai_direct",
+    "ai_with_user_acceptance",
+    "needs_human_judgement",
+    "blocked",
+  ];
+  return order
+    .map((mode) => capabilities.executionModes.find((item) => item.mode === mode && item.status === "ready"))
+    .find(Boolean) || capabilities.executionModes[0];
+}
+
+function buildHandoffGate(summary: PMFlowExecutionSummary, capabilities?: PMFlowExecutionCapabilitySnapshot | null) {
+  const task =
+    summary.activeDevTasks.find((item) => item.id === summary.execution.taskId) || summary.activeDevTasks[0] || null;
+  const formalTask = isFormalDevTask(task) ? task : null;
+  const repo = task?.code_link?.repo || summary.repoBindings[0] || null;
+  const technicalReviewMode = capabilities?.executionModes.find((item) => item.mode === "ai_with_technical_review");
+  const reviewerMissing =
+    technicalReviewMode?.status === "blocked" || Boolean(capabilities?.warnings.some((item) => item.includes("reviewer_missing")));
+  const pendingConfirmations = Number(capabilities?.personalReadiness.waitingForUser.pendingConfirmations ?? 0);
+  const prompt = summary.promptPackage;
+
+  if (!task) {
+    return {
+      task,
+      formalTask,
+      repo,
+      prompt,
+      canCreate: false,
+      actionLabel: "缺少正式任务",
+      reason: "暂无正式开发任务，先从项目记忆形成可执行任务。",
+      mode: primaryExecutionMode(capabilities),
+      requiresTechnicalReview: false,
+    };
+  }
+  if (!formalTask) {
+    return {
+      task,
+      formalTask,
+      repo,
+      prompt,
+      canCreate: false,
+      actionLabel: "等待写入项目记忆",
+      reason: "当前任务仍是候选或草稿，需要先在待确认中心写入项目记忆。",
+      mode: primaryExecutionMode(capabilities),
+      requiresTechnicalReview: false,
+    };
+  }
+  if (reviewerMissing) {
+    return {
+      task,
+      formalTask,
+      repo,
+      prompt,
+      canCreate: false,
+      actionLabel: "先补审查责任",
+      reason: "当前执行模式需要技术审查责任，但 reviewer 尚未明确；请先回到团队层或待确认中心补齐审查责任。",
+      mode: technicalReviewMode || primaryExecutionMode(capabilities),
+      requiresTechnicalReview: true,
+    };
+  }
+  if (pendingConfirmations > 0) {
+    return {
+      task,
+      formalTask,
+      repo,
+      prompt,
+      canCreate: false,
+      actionLabel: "先处理待确认",
+      reason: "当前项目仍有 AI 候选变更待确认，先完成 Change Review，避免执行包基于未确认范围扩散。",
+      mode: primaryExecutionMode(capabilities),
+      requiresTechnicalReview: false,
+    };
+  }
+  if (!repo) {
+    return {
+      task,
+      formalTask,
+      repo,
+      prompt,
+      canCreate: true,
+      actionLabel: "生成交接草案",
+      reason: "可以生成执行交接草案，但仓库落点仍需补齐；这不会声称已经推送到真实代码环境。",
+      mode: primaryExecutionMode(capabilities),
+      requiresTechnicalReview: technicalReviewMode?.status === "ready",
+    };
+  }
+  if (!prompt) {
+    return {
+      task,
+      formalTask,
+      repo,
+      prompt,
+      canCreate: true,
+      actionLabel: "生成上下文包",
+      reason: "正式任务与仓库已就绪，下一步生成开发上下文包供运行端领取。",
+      mode: primaryExecutionMode(capabilities),
+      requiresTechnicalReview: technicalReviewMode?.status === "ready",
+    };
+  }
+  return {
+    task,
+    formalTask,
+    repo,
+    prompt,
+    canCreate: true,
+    actionLabel: "发起执行交接",
+    reason: "复用当前上下文包，并记录新的执行交接起点；代码执行仍在外部运行端完成。",
+    mode: primaryExecutionMode(capabilities),
+    requiresTechnicalReview: technicalReviewMode?.status === "ready",
+  };
+}
+
 export default function ExecutionBridgePage() {
   const {
     state,
@@ -103,7 +226,11 @@ export default function ExecutionBridgePage() {
     executionSummary,
     executionSummaryStatus,
     executionSummaryError,
+    executionCapabilities,
+    executionCapabilitiesStatus,
+    executionCapabilitiesError,
     refetchExecutionSummary,
+    refetchExecutionCapabilities,
     createExecutionHandoff,
     consumeExecutionHandoff,
     recordExecutionProgress,
@@ -111,6 +238,8 @@ export default function ExecutionBridgePage() {
   } = usePMFlowOsStore();
   const demoSummary = dataSourceMode === "demo" ? createDemoExecutionSummary(state) : null;
   const summary = executionSummary || demoSummary;
+  const demoCapabilities = dataSourceMode === "demo" && summary ? createDemoExecutionCapabilities(summary) : null;
+  const capabilities = executionCapabilities || demoCapabilities;
   const isDemo = dataSourceMode === "demo";
   const isLoading = dataSourceMode === "api" && executionSummaryStatus === "loading";
   const error = dataSourceMode === "api" ? executionSummaryError : null;
@@ -132,10 +261,13 @@ export default function ExecutionBridgePage() {
           <button
             type="button"
             className="inline-flex items-center gap-2 rounded-lg border border-[#dfe7f3] bg-white px-3 py-2 text-sm font-semibold text-[#2463EB] transition hover:bg-[#f5f8ff]"
-            onClick={refetchExecutionSummary}
+            onClick={() => {
+              refetchExecutionSummary();
+              refetchExecutionCapabilities();
+            }}
           >
             <RefreshCw className="h-4 w-4" />
-            刷新执行摘要
+            刷新执行态势
           </button>
         ) : null}
       </header>
@@ -154,12 +286,19 @@ export default function ExecutionBridgePage() {
 
       {summary ? (
         <>
+          <ExecutionCapabilityPanel
+            capabilities={capabilities}
+            status={dataSourceMode === "api" ? executionCapabilitiesStatus : "success"}
+            error={dataSourceMode === "api" ? executionCapabilitiesError : null}
+          />
           <ExecutionPipelinePanel summary={summary} />
           <ExecutionWarningList warnings={summary.warnings} />
+          <ExecutionResultSummaryPanel summary={summary} capabilities={capabilities} />
 
           <div className="grid gap-4 min-[1440px]:grid-cols-[360px_minmax(0,1fr)_340px]">
             <ContextPackagePanel
               summary={summary}
+              capabilities={capabilities}
               onCreateHandoff={isDemo ? undefined : createExecutionHandoff}
               onConsumeHandoff={isDemo ? undefined : consumeExecutionHandoff}
               onDemoPush={isDemo ? advanceExecution : undefined}
@@ -168,7 +307,7 @@ export default function ExecutionBridgePage() {
               summary={summary}
               onRecordProgress={isDemo ? undefined : recordExecutionProgress}
             />
-            <AICopilotReviewPanel summary={summary} />
+            <AICopilotReviewPanel summary={summary} capabilities={capabilities} />
           </div>
 
           <ExecutionProgressPanel summary={summary} />
@@ -178,6 +317,127 @@ export default function ExecutionBridgePage() {
   );
 }
 
+function ExecutionCapabilityPanel({
+  capabilities,
+  status,
+  error,
+}: {
+  capabilities?: PMFlowExecutionCapabilitySnapshot | null;
+  status: "idle" | "loading" | "success" | "error";
+  error?: string | null;
+}) {
+  if (status === "loading") {
+    return <ExecutionEmptyState title="正在读取执行能力图谱" summary="执行桥正在同步 AI、工具、人类审查和容量信号，不会先用静态技能表冒充判断。" />;
+  }
+  if (status === "error") {
+    return <ExecutionEmptyState title="执行能力图谱同步失败" summary={error || "执行能力信号暂未同步，可先查看执行摘要和回写证据。"} />;
+  }
+  if (!capabilities) {
+    return <ExecutionEmptyState title="缺少执行能力图谱" summary="当前没有 execution-capabilities 快照，执行桥只展示执行摘要，不做模式裁决。" />;
+  }
+  const primary = primaryExecutionMode(capabilities);
+  const readiness = capabilities.personalReadiness;
+  return (
+    <section className="rounded-lg border border-[#dbe6ff] bg-white px-5 py-4 shadow-[0_1px_2px_rgba(15,23,42,0.03)]">
+      <div className="flex flex-wrap items-start justify-between gap-4">
+        <div>
+          <div className="flex items-center gap-2 text-base font-semibold text-slate-900">
+            <ShieldCheck className="h-4 w-4 text-[#2463EB]" />
+            执行模式判定
+          </div>
+          <p className="mt-1 text-sm leading-6 text-slate-500">
+            这里把正式任务、工具连接、审查责任和容量信号收束成执行模式；缺条件时只能阻塞或生成草案。
+          </p>
+        </div>
+        <StatusChip label={primary?.label || "未判定"} tone={modeTone(primary?.status)} />
+      </div>
+
+      <div className="mt-4 grid gap-3 lg:grid-cols-[minmax(0,1.35fr)_minmax(280px,0.65fr)]">
+        <div className="grid gap-3 md:grid-cols-2 xl:grid-cols-5">
+          {capabilities.executionModes.map((mode) => (
+            <ExecutionModeCard key={mode.mode} mode={mode} active={mode.mode === primary?.mode} />
+          ))}
+        </div>
+        <aside className="rounded-lg border border-[#e8eef8] bg-[#fbfdff] p-4">
+          <div className="flex items-center justify-between gap-3">
+            <h3 className="text-sm font-semibold text-slate-900">我的执行准备度</h3>
+            <StatusChip label={readinessLabel(readiness.status)} tone={modeTone(readiness.status)} />
+          </div>
+          <p className="mt-2 text-sm leading-6 text-slate-600">{readiness.recommendedNextStep}</p>
+          <div className="mt-3 grid grid-cols-3 gap-2">
+            <ExecutionMetric label="待确认" value={`${readiness.waitingForUser.pendingConfirmations ?? 0}`} />
+            <ExecutionMetric label="待我审" value={`${readiness.waitingForUser.pendingReviews ?? 0}`} />
+            <ExecutionMetric label="待验收" value={`${readiness.waitingForUser.waitingAcceptance ?? 0}`} />
+          </div>
+          {readiness.blockers.length ? (
+            <ul className="mt-3 space-y-1.5">
+              {readiness.blockers.slice(0, 3).map((blocker) => (
+                <li key={blocker} className="text-xs leading-5 text-amber-700">
+                  {blocker}
+                </li>
+              ))}
+            </ul>
+          ) : null}
+        </aside>
+      </div>
+    </section>
+  );
+}
+
+function ExecutionModeCard({ mode, active }: { mode: PMFlowExecutionModeCapability; active: boolean }) {
+  return (
+    <article
+      className={cn(
+        "rounded-lg border p-3",
+        active ? "border-[#bfd4ff] bg-[#f5f8ff]" : "border-[#e8eef8] bg-white",
+        mode.status === "blocked" && "border-rose-200 bg-rose-50"
+      )}
+    >
+      <div className="flex items-center justify-between gap-2">
+        <StatusChip label={readinessLabel(mode.status)} tone={modeTone(mode.status)} />
+        {active ? <span className="text-[11px] font-semibold text-[#2463EB]">当前建议</span> : null}
+      </div>
+      <div className="mt-3 text-sm font-semibold text-slate-900">{mode.label}</div>
+      <p className="mt-1 text-xs leading-5 text-slate-500">{mode.reason}</p>
+    </article>
+  );
+}
+
+function ExecutionResultSummaryPanel({
+  summary,
+  capabilities,
+}: {
+  summary: PMFlowExecutionSummary;
+  capabilities?: PMFlowExecutionCapabilitySnapshot | null;
+}) {
+  const result = deriveExecutionResult(summary, capabilities);
+  return (
+    <section className="rounded-lg border border-[#dbe6ff] bg-white px-5 py-4 shadow-[0_1px_2px_rgba(15,23,42,0.03)]">
+      <div className="flex flex-wrap items-start justify-between gap-4">
+        <div>
+          <div className="flex items-center gap-2 text-base font-semibold text-slate-900">
+            <ClipboardCheck className="h-4 w-4 text-[#2463EB]" />
+            本次执行结果
+          </div>
+          <p className="mt-1 text-sm leading-6 text-slate-500">
+            汇总运行端、PR、测试、阻塞和验收信号；缺证据时停在待回写，不进入发布成功表达。
+          </p>
+        </div>
+        <StatusChip label={result.statusLabel} tone={result.tone} />
+      </div>
+      <div className="mt-4 grid gap-3 md:grid-cols-2 xl:grid-cols-4">
+        <ExecutionMetric label="运行端" value={result.runtimeLabel} />
+        <ExecutionMetric label="代码证据" value={result.codeEvidenceLabel} />
+        <ExecutionMetric label="测试证据" value={result.testEvidenceLabel} tone={result.testTone === "risk" ? "warning" : "default"} />
+        <ExecutionMetric label="下一步" value={result.nextAction} tone={result.tone === "risk" ? "warning" : "default"} />
+      </div>
+      {result.blocker ? (
+        <div className="mt-4 rounded-lg border border-rose-200 bg-rose-50 px-4 py-3 text-sm leading-6 text-rose-700">{result.blocker}</div>
+      ) : null}
+    </section>
+  );
+}
+
 function ExecutionPipelinePanel({ summary }: { summary: PMFlowExecutionSummary }) {
   const counts = summary.devSync;
   const steps = buildPipelineSteps(summary);
@@ -262,11 +522,13 @@ function PipelineStepCard({
 
 function ContextPackagePanel({
   summary,
+  capabilities,
   onCreateHandoff,
   onConsumeHandoff,
   onDemoPush,
 }: {
   summary: PMFlowExecutionSummary;
+  capabilities?: PMFlowExecutionCapabilitySnapshot | null;
   onCreateHandoff?: (payload: {
     task_id: string;
     repo_binding_id?: string | null;
@@ -277,20 +539,12 @@ function ContextPackagePanel({
   onDemoPush?: () => void;
 }) {
   const prompt = summary.promptPackage;
-  const handoffTask =
+  const handoffGate = buildHandoffGate(summary, capabilities);
-    summary.activeDevTasks.find((task) => task.id === summary.execution.taskId) || summary.activeDevTasks[0] || null;
+  const handoffTask = handoffGate.task;
-  const formalHandoffTask = isFormalDevTask(handoffTask) ? handoffTask : null;
+  const formalHandoffTask = handoffGate.formalTask;
-  const handoffRepo = handoffTask?.code_link?.repo || summary.repoBindings[0] || null;
+  const handoffRepo = handoffGate.repo;
-  const canCreateHandoff = Boolean(formalHandoffTask);
+  const canCreateHandoff = handoffGate.canCreate;
-  const handoffNextStep = !handoffTask
+  const handoffNextStep = handoffGate.reason;
-    ? "暂无正式开发任务，先从项目记忆生成可执行任务。"
-    : !formalHandoffTask
-      ? "当前任务仍是候选或草稿，需要先在待确认中心写入项目记忆。"
-      : !handoffRepo
-        ? "可以生成交接草案，但仓库落点仍需补齐。"
-        : !prompt
-          ? "生成开发上下文包后，运行端才能领取任务。"
-          : "复用当前上下文包，并记录新的执行交接起点。";
 
   return (
     <aside className="rounded-lg border border-[#e2e9f5] bg-white shadow-[0_1px_2px_rgba(15,23,42,0.03)]">
@@ -350,8 +604,14 @@ function ContextPackagePanel({
                   <FieldRow label="任务" value={handoffTask?.title || "缺少正式任务"} />
                   <FieldRow label="状态" value={taskStatusLabel(handoffTask?.status || "missing")} />
                   <FieldRow label="仓库" value={handoffRepo?.full_name || "未绑定仓库"} />
-                  <FieldRow label="运行端" value="本地桥" />
+                  <FieldRow label="执行模式" value={handoffGate.mode?.label || "未同步"} />
+                  <FieldRow label="运行端" value="本地桥 / Codex / GitHub Agent" />
                 </dl>
+                {handoffGate.requiresTechnicalReview ? (
+                  <div className="mt-3 rounded-lg border border-amber-200 bg-amber-50 px-3 py-2 text-xs leading-5 text-amber-700">
+                    该模式需要技术审查责任，运行端结果回写后仍需 reviewer 或用户验收，不能直接视为完成。
+                  </div>
+                ) : null}
                 <button
                   type="button"
                   className={cn(
@@ -362,7 +622,7 @@ function ContextPackagePanel({
                   )}
                   disabled={!canCreateHandoff}
                   onClick={() => {
-                    if (!formalHandoffTask) return;
+                    if (!formalHandoffTask || !canCreateHandoff) return;
                     onCreateHandoff({
                       task_id: formalHandoffTask.id,
                       repo_binding_id: handoffRepo?.id || null,
@@ -372,7 +632,7 @@ function ContextPackagePanel({
                   }}
                 >
                   <Rocket className="h-4 w-4" />
-                  {handoffRepo ? "发起执行交接" : "生成交接草案"}
+                  {handoffGate.actionLabel}
                 </button>
               </div>
             </div>
@@ -769,9 +1029,16 @@ function WritebackInput({
   );
 }
 
-function AICopilotReviewPanel({ summary }: { summary: PMFlowExecutionSummary }) {
+function AICopilotReviewPanel({
-  const insights = buildReviewInsights(summary);
+  summary,
-  const nextActions = buildNextActions(summary);
+  capabilities,
+}: {
+  summary: PMFlowExecutionSummary;
+  capabilities?: PMFlowExecutionCapabilitySnapshot | null;
+}) {
+  const insights = buildReviewInsights(summary, capabilities);
+  const nextActions = buildNextActions(summary, capabilities);
+  const result = deriveExecutionResult(summary, capabilities);
   return (
     <aside className="rounded-lg border border-[#e2e9f5] bg-white shadow-[0_1px_2px_rgba(15,23,42,0.03)]">
       <div className="flex items-center justify-between border-b border-[#edf2f8] px-5 py-4">
@@ -783,7 +1050,7 @@ function AICopilotReviewPanel({ summary }: { summary: PMFlowExecutionSummary })
       </div>
       <div className="space-y-4 p-5">
         <section className="rounded-lg border border-[#e8eef8] bg-[#fbfdff] p-4">
-          <h3 className="text-sm font-semibold text-slate-900">审查重点</h3>
+          <h3 className="text-sm font-semibold text-slate-900">执行裁决重点</h3>
           <div className="mt-3 space-y-3">
             {insights.map((item) => (
               <div key={item.label} className="flex items-start justify-between gap-3">
@@ -799,7 +1066,10 @@ function AICopilotReviewPanel({ summary }: { summary: PMFlowExecutionSummary })
 
         <section className="rounded-lg border border-[#e8eef8] bg-white p-4">
           <h3 className="text-sm font-semibold text-slate-900">风险结论</h3>
-          <p className="mt-2 text-sm leading-6 text-slate-500">{buildRiskConclusion(summary)}</p>
+          <p className="mt-2 text-sm leading-6 text-slate-500">{buildRiskConclusion(summary, capabilities)}</p>
+          <div className="mt-3 rounded-lg border border-[#e8eef8] bg-[#fbfdff] px-3 py-2 text-xs leading-5 text-slate-500">
+            当前结果：{result.statusLabel}；下一步：{result.nextAction}
+          </div>
         </section>
 
         <section className="rounded-lg border border-[#e8eef8] bg-white p-4">
@@ -1012,21 +1282,30 @@ function buildPipelineSteps(summary: PMFlowExecutionSummary) {
   }));
 }
 
-function buildReviewInsights(summary: PMFlowExecutionSummary) {
+function buildReviewInsights(summary: PMFlowExecutionSummary, capabilities?: PMFlowExecutionCapabilitySnapshot | null) {
   const hasPrompt = Boolean(summary.promptPackage);
   const hasTask = summary.activeDevTasks.length > 0;
   const hasTest = summary.recentProgressEvents.some((event) => String(event.event_type).includes("test"));
   const hasFailure = summary.execution.status === "failed" || summary.recentProgressEvents.some((event) => ["test_failed", "failed", "blocked"].includes(String(event.event_type)));
   const hasPr = summary.recentProgressEvents.some((event) => String(event.event_type).includes("pr") || event.pr_number);
+  const primary = primaryExecutionMode(capabilities);
+  const technicalReviewMode = capabilities?.executionModes.find((item) => item.mode === "ai_with_technical_review");
+  const waitingAcceptance = Number(capabilities?.personalReadiness.waitingForUser.waitingAcceptance ?? 0);
   return [
     {
-      label: "需求覆盖",
+      label: "执行模式",
-      summary: hasPrompt && hasTask ? "任务、上下文包和项目记忆已形成可追溯关系。" : "还缺正式任务或上下文包，暂不能判断需求覆盖。",
+      summary: primary ? `${primary.label}：${primary.reason}` : "执行能力图谱未同步，只能依据执行摘要判断。",
+      status: primary ? readinessLabel(primary.status) : "未同步",
+      tone: primary ? modeTone(primary.status) : "muted" as const,
+    },
+    {
+      label: "上下文覆盖",
+      summary: hasPrompt && hasTask ? "正式任务、上下文包和项目记忆已形成可追溯关系。" : "还缺正式任务或上下文包，暂不能判断需求覆盖。",
       status: hasPrompt && hasTask ? "已通过" : "待补齐",
       tone: hasPrompt && hasTask ? "success" as const : "warning" as const,
     },
     {
-      label: "代码质量",
+      label: "代码证据",
       summary: hasPr ? "PR 或提交证据已回流，可进入代码审查。" : "暂无 PR 或提交证据，代码质量只能等待运行端回写。",
       status: hasPr ? "可审查" : "待回写",
       tone: hasPr ? "brand" as const : "muted" as const,
@@ -1043,17 +1322,32 @@ function buildReviewInsights(summary: PMFlowExecutionSummary) {
       status: summary.recentProgressEvents.length ? "已回写" : "等待中",
       tone: summary.recentProgressEvents.length ? "success" as const : "muted" as const,
     },
+    {
+      label: "验收责任",
+      summary:
+        technicalReviewMode?.status === "ready"
+          ? "当前更适合 AI 执行后绑定技术审查，不能把 agent 结果直接视为完成。"
+          : waitingAcceptance > 0
+            ? "已有执行结果等待用户验收，先处理验收再进入发布门禁。"
+            : "暂未出现明确技术审查或用户验收阻塞。",
+      status: technicalReviewMode?.status === "ready" || waitingAcceptance > 0 ? "需处理" : "正常",
+      tone: technicalReviewMode?.status === "ready" || waitingAcceptance > 0 ? "warning" as const : "success" as const,
+    },
   ];
 }
 
-function buildRiskConclusion(summary: PMFlowExecutionSummary) {
+function buildRiskConclusion(summary: PMFlowExecutionSummary, capabilities?: PMFlowExecutionCapabilitySnapshot | null) {
+  const technicalReviewMode = capabilities?.executionModes.find((item) => item.mode === "ai_with_technical_review");
+  if (technicalReviewMode?.status === "blocked") return "当前执行需要技术审查责任，但 reviewer 尚未明确，不能发起真实执行交接。";
   if (summary.execution.status === "failed") return "当前执行链路存在失败事件，发布门禁应保持阻塞，先定位失败原因并补充测试证据。";
   if (summary.warnings.length) return "当前主要风险来自执行准备缺口，补齐正式任务、仓库、上下文包和进展回写后再推进到 PR 或发布判断。";
   if (summary.execution.status === "synced") return "执行结果已回写，可进入验收或发布门禁判断，但仍需要保留 PR、测试和回滚证据。";
   return "执行链路正在推进，建议持续观察运行端回写，并在 PR 与测试证据完整后再进入发布候选。";
 }
 
-function buildNextActions(summary: PMFlowExecutionSummary) {
+function buildNextActions(summary: PMFlowExecutionSummary, capabilities?: PMFlowExecutionCapabilitySnapshot | null) {
+  const gate = buildHandoffGate(summary, capabilities);
+  if (!gate.canCreate && gate.actionLabel !== "发起执行交接") return [gate.reason, "按执行模式判定先补齐缺口，再回到执行桥继续交接。"];
   if (!summary.activeDevTasks.length) return ["先从项目记忆形成正式开发任务。", "回到项目工作台确认任务来源和验收标准。"];
   if (!summary.repoBindings.length) return ["绑定代码仓库或确认 GitHub App 连接状态。", "仓库就绪后重新生成执行交接包。"];
   if (!summary.promptPackage) return ["生成开发上下文包。", "确认上下文包包含需求、约束、验收标准和风险。"];
@@ -1062,6 +1356,96 @@ function buildNextActions(summary: PMFlowExecutionSummary) {
   return ["继续观察 PR、测试和验收回写。", "证据完整后进入发布与增长门禁。"];
 }
 
+function deriveExecutionResult(summary: PMFlowExecutionSummary, capabilities?: PMFlowExecutionCapabilitySnapshot | null) {
+  const events = summary.recentProgressEvents;
+  const hasFailure = summary.execution.status === "failed" || events.some((event) => ["test_failed", "failed", "blocked"].includes(String(event.event_type)));
+  const hasAccepted = events.some((event) => String(event.event_type) === "accepted");
+  const hasTestPassed = events.some((event) => ["test_passed", "checks_passed"].includes(String(event.event_type)));
+  const hasPr = events.some((event) => String(event.event_type).includes("pr") || event.pr_number);
+  const hasCommit = events.some((event) => String(event.event_type).includes("commit") || event.commit_sha);
+  const lastEvent = events[0];
+  const runtimeLabel = sourceLabel(lastEvent?.source) || "等待运行端领取";
+  const technicalReviewMode = capabilities?.executionModes.find((item) => item.mode === "ai_with_technical_review");
+  const needsTechnicalReview = technicalReviewMode?.status === "ready";
+  const blocker = events.find((event) => ["test_failed", "failed", "blocked"].includes(String(event.event_type)))?.summary || null;
+
+  if (hasFailure) {
+    return {
+      statusLabel: "执行受阻",
+      tone: "risk" as const,
+      runtimeLabel,
+      codeEvidenceLabel: hasPr ? "PR 已回流" : hasCommit ? "提交已回流" : "缺代码证据",
+      testEvidenceLabel: summary.execution.testSummary && summary.execution.testSummary !== "pending" ? normalizePending(summary.execution.testSummary) : "失败或阻塞",
+      testTone: "risk" as const,
+      nextAction: "修复后回写",
+      blocker: blocker || "存在失败或阻塞事件，需要先补充原因、影响和下一步。",
+    };
+  }
+  if (hasAccepted) {
+    return {
+      statusLabel: "已验收",
+      tone: "success" as const,
+      runtimeLabel,
+      codeEvidenceLabel: hasPr ? "PR 已回流" : hasCommit ? "提交已回流" : "缺代码证据",
+      testEvidenceLabel: hasTestPassed ? "测试通过" : "测试证据不足",
+      testTone: hasTestPassed ? "success" as const : "warning" as const,
+      nextAction: "进入发布门禁",
+      blocker: null,
+    };
+  }
+  if (hasTestPassed || summary.execution.status === "synced") {
+    return {
+      statusLabel: needsTechnicalReview ? "待技术审查" : "待用户验收",
+      tone: "warning" as const,
+      runtimeLabel,
+      codeEvidenceLabel: hasPr ? "PR 已回流" : hasCommit ? "提交已回流" : "缺代码证据",
+      testEvidenceLabel: hasTestPassed ? "测试通过" : normalizePending(summary.execution.testSummary),
+      testTone: hasTestPassed ? "success" as const : "warning" as const,
+      nextAction: needsTechnicalReview ? "技术审查" : "用户验收",
+      blocker: null,
+    };
+  }
+  if (events.length) {
+    return {
+      statusLabel: "执行中",
+      tone: "brand" as const,
+      runtimeLabel,
+      codeEvidenceLabel: hasPr ? "PR 已回流" : hasCommit ? "提交已回流" : "等待 PR/提交",
+      testEvidenceLabel: "等待测试",
+      testTone: "warning" as const,
+      nextAction: "继续回写证据",
+      blocker: null,
+    };
+  }
+  return {
+    statusLabel: "待回写",
+    tone: "muted" as const,
+    runtimeLabel: "未领取",
+    codeEvidenceLabel: "缺代码证据",
+    testEvidenceLabel: "缺测试证据",
+    testTone: "warning" as const,
+    nextAction: "发起交接",
+    blocker: null,
+  };
+}
+
+function readinessLabel(status?: CapabilityTone | string | null) {
+  const map: Record<string, string> = {
+    ready: "就绪",
+    partial: "待补齐",
+    blocked: "阻塞",
+    unknown: "未知",
+  };
+  return status ? map[status] || status : "未知";
+}
+
+function modeTone(status?: CapabilityTone | string | null): "brand" | "success" | "warning" | "risk" | "muted" {
+  if (status === "ready") return "success";
+  if (status === "partial") return "warning";
+  if (status === "blocked") return "risk";
+  return "muted";
+}
+
 function eventIcon(eventType: string) {
   const lower = eventType.toLowerCase();
   if (lower.includes("pr") || lower.includes("pull")) return <GitPullRequest className="h-4 w-4 text-[#2463EB]" />;
@@ -1329,3 +1713,79 @@ function createDemoExecutionSummary(state: ReturnType<typeof usePMFlowOsStore>["
     warnings: [],
   };
 }
+
+function createDemoExecutionCapabilities(summary: PMFlowExecutionSummary): PMFlowExecutionCapabilitySnapshot {
+  const hasRepo = summary.repoBindings.length > 0;
+  const hasPrompt = Boolean(summary.promptPackage);
+  const hasProgress = summary.recentProgressEvents.length > 0;
+  return {
+    projectId: summary.projectId,
+    executionModes: [
+      {
+        mode: "ai_direct",
+        label: "AI 直接执行",
+        status: hasRepo && hasPrompt ? "ready" : "partial",
+        reason: "演示模式下仅展示执行模式，不触发真实运行端。",
+        targetHref: `/execution-bridge?projectId=${summary.projectId}`,
+      },
+      {
+        mode: "ai_with_user_acceptance",
+        label: "AI 执行 + 用户验收",
+        status: hasRepo ? "ready" : "blocked",
+        reason: "适合低风险改动，执行结果仍需要用户验收。",
+        targetHref: `/execution-bridge?projectId=${summary.projectId}`,
+      },
+      {
+        mode: "ai_with_technical_review",
+        label: "AI 执行 + 技术审查",
+        status: "partial",
+        reason: "涉及架构、权限、数据或发布门禁时需要技术审查。",
+        targetHref: `/team-layer?projectId=${summary.projectId}`,
+      },
+      {
+        mode: "needs_human_judgement",
+        label: "需要人类先补判断",
+        status: "partial",
+        reason: "目标、范围或验收标准不清时先回到待确认中心。",
+        targetHref: `/change-review?projectId=${summary.projectId}`,
+      },
+      {
+        mode: "blocked",
+        label: "当前不可执行",
+        status: hasRepo && hasPrompt && hasProgress ? "unknown" : "ready",
+        reason: "缺上下文包、运行端领取或回写证据时，必须显式阻塞。",
+        targetHref: `/execution-bridge?projectId=${summary.projectId}`,
+      },
+    ],
+    personalReadiness: {
+      status: hasRepo && hasPrompt ? "ready" : "partial",
+      toolReadiness: [
+        {
+          key: "repo",
+          label: "Repo 连接",
+          status: hasRepo ? "ready" : "blocked",
+          summary: hasRepo ? "演示仓库已就绪。" : "缺少 repo。",
+          href: `/execution-bridge?projectId=${summary.projectId}`,
+        },
+        {
+          key: "prompt_package",
+          label: "执行 Prompt Package",
+          status: hasPrompt ? "ready" : "partial",
+          summary: hasPrompt ? "已有演示上下文包。" : "尚未生成上下文包。",
+          href: `/execution-bridge?projectId=${summary.projectId}`,
+        },
+      ],
+      waitingForUser: {
+        pendingConfirmations: 0,
+        pendingReviews: 0,
+        waitingAcceptance: hasProgress ? 1 : 0,
+      },
+      blockers: hasPrompt ? [] : ["尚未生成上下文包。"],
+      recommendedNextStep: hasPrompt ? "可以推进演示执行状态。" : "先生成或确认执行上下文包。",
+      targetHref: `/execution-bridge?projectId=${summary.projectId}`,
+    },
+    teamSignals: [],
+    adaptiveCapacitySummary: null,
+    warnings: [],
+  };
+}

frontend/src/features/pmflow-os/store.api.test.tsx

@@ -31,6 +31,7 @@ import {
 import { createInitialPMFlowState } from "@/features/pmflow-os/mockData";
 import CommandCenterPage from "@/features/pmflow-os/pages/CommandCenterPage";
 import ChangeReviewPage from "@/features/pmflow-os/pages/ChangeReviewPage";
+import ExecutionBridgePage from "@/features/pmflow-os/pages/ExecutionBridgePage";
 import StructuredExtractionPage from "@/features/pmflow-os/pages/StructuredExtractionPage";
 import TeamLayerPage from "@/features/pmflow-os/pages/TeamLayerPage";
 import WorkGraphPage from "@/features/pmflow-os/pages/WorkGraphPage";
@@ -704,6 +705,72 @@ describe("PMFlowOsProvider real change review actions", () => {
     expect(screen.queryByText(/Team Layer/)).not.toBeInTheDocument();
   });
 
+  it("renders execution capability mode and result summary on execution bridge", async () => {
+    getExecutionCapabilitiesMock.mockResolvedValueOnce({
+      projectId: "123",
+      executionModes: [
+        {
+          mode: "ai_direct",
+          label: "AI 直接执行",
+          status: "ready",
+          reason: "正式任务、仓库和上下文包已就绪。",
+          targetHref: "/execution-bridge?projectId=123",
+        },
+        {
+          mode: "ai_with_user_acceptance",
+          label: "AI 执行 + 用户验收",
+          status: "ready",
+          reason: "执行后需要用户验收。",
+          targetHref: "/execution-bridge?projectId=123",
+        },
+        {
+          mode: "ai_with_technical_review",
+          label: "AI 执行 + 技术审查",
+          status: "partial",
+          reason: "高风险时需要 reviewer。",
+          targetHref: "/team-layer?projectId=123",
+        },
+        {
+          mode: "needs_human_judgement",
+          label: "需要人类先补判断",
+          status: "partial",
+          reason: "范围不清时先回待确认。",
+          targetHref: "/change-review?projectId=123",
+        },
+        {
+          mode: "blocked",
+          label: "当前不可执行",
+          status: "unknown",
+          reason: "缺条件才阻塞。",
+          targetHref: "/execution-bridge?projectId=123",
+        },
+      ],
+      personalReadiness: {
+        status: "ready",
+        toolReadiness: [],
+        waitingForUser: { pendingConfirmations: 0, pendingReviews: 0, waitingAcceptance: 0 },
+        blockers: [],
+        recommendedNextStep: "可以继续生成或验收 AI 执行包。",
+        targetHref: "/execution-bridge?projectId=123",
+      },
+      teamSignals: [],
+      warnings: [],
+    });
+
+    render(
+      <MemoryRouter>
+        <ExecutionBridgePage />
+      </MemoryRouter>,
+      { wrapper: wrapper() }
+    );
+
+    expect(await screen.findByText("执行模式判定")).toBeInTheDocument();
+    expect(screen.getAllByText("AI 直接执行").length).toBeGreaterThan(0);
+    expect(screen.getByText("本次执行结果")).toBeInTheDocument();
+    expect(screen.getAllByText("发起执行交接").length).toBeGreaterThan(0);
+    expect(screen.getByText("正式任务、仓库和上下文包已就绪。")).toBeInTheDocument();
+  });
+
   it("renders Change Review from real projection and does not show demo fallback rows", async () => {
     render(
       <MemoryRouter>
@@ -1198,6 +1265,44 @@ describe("PMFlowOsProvider real change review actions", () => {
     await waitFor(() => expect(result.current.dataSourceError).toMatch(/正式任务/));
   });
 
+  it("blocks execution handoff when technical review responsibility is missing", async () => {
+    getExecutionCapabilitiesMock.mockResolvedValueOnce({
+      projectId: "123",
+      executionModes: [
+        {
+          mode: "ai_with_technical_review",
+          label: "AI 执行 + 技术审查",
+          status: "blocked",
+          reason: "缺少 reviewer。",
+          targetHref: "/team-layer?projectId=123",
+        },
+      ],
+      personalReadiness: {
+        status: "partial",
+        toolReadiness: [],
+        waitingForUser: { pendingConfirmations: 0, pendingReviews: 0, waitingAcceptance: 0 },
+        blockers: [],
+        recommendedNextStep: "先补齐 reviewer。",
+        targetHref: "/team-layer?projectId=123",
+      },
+      teamSignals: [],
+      warnings: ["reviewer_missing: 存在待审 ChangeSet 缺少 reviewer/owner。"],
+    });
+    const { result } = renderHook(() => usePMFlowOsStore(), { wrapper: wrapper() });
+
+    await waitFor(() => expect(result.current.executionCapabilitiesStatus).toBe("success"));
+    act(() =>
+      result.current.createExecutionHandoff({
+        task_id: "task-99",
+        repo_binding_id: "601",
+        runtime_target: "local_bridge",
+      })
+    );
+
+    expect(createExecutionHandoffMock).not.toHaveBeenCalled();
+    await waitFor(() => expect(result.current.dataSourceError).toMatch(/技术审查责任/));
+  });
+
   it("keeps reviewer routing empty when payload omits it", async () => {
     const state = realState();
     delete state.reviewerRouting;

frontend/src/features/pmflow-os/store.tsx

@@ -16,6 +16,7 @@ import {
   createPMFlowThreadMessage,
   deferPMFlowChangeItem,
   editPMFlowChangeItem,
+  getPMFlowExecutionCapabilities,
   getPMFlowExecutionSummary,
   getPMFlowTruthState,
   type PMFlowExecutionHandoffRequest,
@@ -33,6 +34,7 @@ import { createInitialPMFlowState } from "@/features/pmflow-os/mockData";
 import type {
   ChangeSet,
   PMFlowAIFeedbackEvent,
+  PMFlowExecutionCapabilitySnapshot,
   PMFlowExecutionStatus,
   ProjectTruthState,
   Release,
@@ -646,8 +648,12 @@ interface PMFlowOsStoreValue {
   executionSummary: PMFlowExecutionSummary | null;
   executionSummaryStatus: "idle" | "loading" | "success" | "error";
   executionSummaryError: string | null;
+  executionCapabilities: PMFlowExecutionCapabilitySnapshot | null;
+  executionCapabilitiesStatus: "idle" | "loading" | "success" | "error";
+  executionCapabilitiesError: string | null;
   refetchTruthState: () => void;
   refetchExecutionSummary: () => void;
+  refetchExecutionCapabilities: () => void;
   sendMessage: (content: string) => void;
   generateExtraction: (proposedItems?: PMFlowProposedChangeItem[]) => void;
   skipExtractionRound: (summary?: string) => void;
@@ -704,6 +710,11 @@ export function PMFlowOsProvider({ children }: { children: ReactNode }) {
     queryFn: () => getPMFlowExecutionSummary(dataSource.projectId || ""),
     enabled: dataSource.mode === "api" && Boolean(dataSource.projectId),
   });
+  const executionCapabilitiesQuery = useQuery({
+    queryKey: ["pmflow-os", "execution-capabilities", dataSource.projectId],
+    queryFn: () => getPMFlowExecutionCapabilities(dataSource.projectId || ""),
+    enabled: dataSource.mode === "api" && Boolean(dataSource.projectId),
+  });
 
   useEffect(() => {
     setReadError(null);
@@ -763,8 +774,10 @@ export function PMFlowOsProvider({ children }: { children: ReactNode }) {
     if (dataSource.mode !== "api") return;
     void queryClient.invalidateQueries({ queryKey: ["pmflow-os", "truth-state", dataSource.projectId] });
     void queryClient.invalidateQueries({ queryKey: ["pmflow-os", "execution-summary", dataSource.projectId] });
+    void queryClient.invalidateQueries({ queryKey: ["pmflow-os", "execution-capabilities", dataSource.projectId] });
     void truthStateQuery.refetch();
     void executionSummaryQuery.refetch();
+    void executionCapabilitiesQuery.refetch();
   };
   const runApiAction = (
     action: () => Promise<unknown> | unknown,
@@ -791,7 +804,7 @@ export function PMFlowOsProvider({ children }: { children: ReactNode }) {
         dataSourceStatus:
           dataSource.mode === "empty"
             ? "idle"
-            : truthStateQuery.isLoading || truthStateQuery.isFetching || executionSummaryQuery.isFetching
+            : truthStateQuery.isLoading || truthStateQuery.isFetching || executionSummaryQuery.isFetching || executionCapabilitiesQuery.isFetching
               ? "loading"
               : readError
                 ? "error"
@@ -816,7 +829,26 @@ export function PMFlowOsProvider({ children }: { children: ReactNode }) {
             : executionSummaryQuery.error instanceof Error
               ? executionSummaryQuery.error.message
               : executionSummaryQuery.error
-                ? "PMFlow OS 执行摘要读取失败"
+              ? "PMFlow OS 执行摘要读取失败"
+              : null,
+        executionCapabilities: dataSource.mode === "api" ? executionCapabilitiesQuery.data ?? null : null,
+        executionCapabilitiesStatus:
+          dataSource.mode === "empty"
+            ? "idle"
+            : dataSource.mode === "demo"
+              ? "success"
+              : executionCapabilitiesQuery.isLoading
+                ? "loading"
+                : executionCapabilitiesQuery.error
+                  ? "error"
+                  : "success",
+        executionCapabilitiesError:
+          dataSource.mode !== "api"
+            ? null
+            : executionCapabilitiesQuery.error instanceof Error
+              ? executionCapabilitiesQuery.error.message
+              : executionCapabilitiesQuery.error
+                ? "PMFlow OS 执行能力读取失败"
                 : null,
         refetchTruthState: () => {
           if (dataSource.mode === "api") refetchApiState();
@@ -826,6 +858,11 @@ export function PMFlowOsProvider({ children }: { children: ReactNode }) {
           void queryClient.invalidateQueries({ queryKey: ["pmflow-os", "execution-summary", dataSource.projectId] });
           void executionSummaryQuery.refetch();
         },
+        refetchExecutionCapabilities: () => {
+          if (dataSource.mode !== "api") return;
+          void queryClient.invalidateQueries({ queryKey: ["pmflow-os", "execution-capabilities", dataSource.projectId] });
+          void executionCapabilitiesQuery.refetch();
+        },
         sendMessage: (content) => {
           if (dataSource.mode === "api" && dataSource.projectId) {
             setReadError(null);
@@ -988,6 +1025,13 @@ export function PMFlowOsProvider({ children }: { children: ReactNode }) {
               setReadError("执行交接只能针对已写入项目记忆后的正式任务；候选任务或待确认变更不能直接分派执行。");
               return;
             }
+            const capability = executionCapabilitiesQuery.data;
+            const technicalReviewMode = capability?.executionModes.find((item) => item.mode === "ai_with_technical_review");
+            const reviewerMissing = technicalReviewMode?.status === "blocked" || capability?.warnings.some((item) => item.includes("reviewer_missing"));
+            if (reviewerMissing) {
+              setReadError("当前执行模式需要技术审查责任，但 reviewer 尚未明确；请先回到团队层或待确认中心补齐审查责任。");
+              return;
+            }
             runApiAction(() => createPMFlowExecutionHandoff(dataSource.projectId || "", payload));
             return;
           }