migration: close candidate maintenance audit

2026-07-31 21:24:37 +08:00
parent faac60b1a6
commit 406118bba6
17 changed files with 844 additions and 120 deletions
@@ -1,40 +1,70 @@
-# Architecture
+# Candidate architecture

-The normative governance contract is documented in
-`docs/governance/architecture-standard.md`. This file describes the currently deployed
-shape; the standard defines the target boundaries and the rules applied during migration.
+`app/` is the behavior-preserving modular candidate. The original `webapp/` runtime remains
+the product and visual baseline until manual acceptance. `next/` is a rejected, frozen
+implementation and is not a source for this directory.

-The application intentionally keeps a small deployment footprint: one Python process, one
-SQLite database, and a build-free browser client. The internal boundaries are nevertheless
-explicit so new features do not bypass account isolation or data-quality rules.
+The application deliberately remains a modular monolith: one Python process, one SQLite WAL
+database, and a build-free HTML/CSS/JavaScript client. The migration changed source ownership
+and imports, not the technology stack or observable product behavior.

-## Backend boundaries
+## Runtime path

- `server.py`: application services and HTTP request/response wiring.
- `api_access.py`: the single authorization policy for authenticated, member, and admin APIs.
- `app_config.py`: runtime paths, local environment loading, and shared input validation.
- `database.py`: SQLite schema, migrations, and persistence operations.
- `tushare_client.py` and `realtime_aggregator.py`: external market-data adapters.
- `sentiment_engine.py`, `screener.py`, and `heaven_engine.py`: deterministic domain logic.
- `mentor_agent.py`, `heaven_agent.py`, and `llm_strategy.py`: bounded LLM adapters.
+```text
+browser
+  -> frontend/shared/api.js
+  -> backend HTTP transport and feature HTTP mixins
+  -> feature services
+  -> repositories / DataGateway / LLMGateway
+  -> SQLite / market providers / model providers

-## Data ownership
+background scheduler
+  -> backend/jobs
+  -> the same feature services and repositories
+```

-Public market snapshots, stock factors, built-in strategies, limit-up reasons, seat aliases,
-and sector-element mappings are shared. Only administrators can modify shared knowledge.
+## Source ownership

-Watchlists, review notes, custom strategies, screener runs, mentor conversations, birth data,
-alerts, trading journals, and assistant conversations are owned by a user ID and must be
-queried with that ID. LLM features additionally require active membership.
+- `server.py` is the stable command/import facade. Runtime composition lives in
+  `backend/application.py` and `backend/bootstrap/`.
+- `backend/http/` owns common authentication, request IDs, responses, static delivery, and
+  error normalization. Feature-specific transport handlers live beside their feature.
+- `backend/features/<feature>/` owns the mechanically moved service, repository, HTTP, agent,
+  or deterministic calculation code for that product area.
+- `backend/data/` owns provider construction, source policy, provenance, units, freshness,
+  coverage, and display-versus-calculation eligibility.
+- `backend/database/` owns connection management, ordered migrations, and narrow repository
+  adapters. Root `database.py` remains the legacy schema/composition anchor and combines the
+  feature repository mixins; do not add feature queries to it.
+- `backend/jobs/` owns job definitions, locks, retries, idempotency, and persisted run state.
+- `backend/llm/` owns model selection, membership/quota checks, fallback, streaming rules,
+  and call audit. Feature agents only prepare context and provider payloads.
+- `frontend/shared/` is the only browser API/state/Shell/component boundary.
+- `frontend/pages/` owns page-local behavior. The original runtime was split mechanically;
+  source markers and preservation tests prove that the pieces reassemble to the audited
+  original, apart from explicitly registered trial retirements.
+- `frontend/styles/`, `frontend/shared/tokens.css`, and the Wentian page stylesheet preserve
+  the approved cascade and light/dark/mobile behavior.
+- `config/` is the versioned registry for pages, features, APIs, datasets, quality rules,
+  jobs, and the generated candidate architecture inventory.

-## Data integrity
+Root modules such as `screener.py`, `tushare_client.py`, and `mentor_agent.py` are compatibility
+aliases to canonical modules. They contain no second implementation and remain only because
+the original public import surface is part of the preservation contract.

-Production reads never synthesize market prices. A failed live request may use the latest real
-snapshot at or before the requested date. When no real snapshot exists, the API reports that
-the data is unavailable. Demo builders remain test fixtures only.
+## Non-negotiable maintenance rules

-## Change contract
+1. Preserve account ownership in every user-private query and test it with two accounts.
+2. Browser requests go through `frontend/shared/api.js`; provider calls go through the data
+   boundary; model calls go through `backend/llm/`.
+3. Calculation datasets fail closed when required source, date, unit, freshness, or coverage
+   evidence is missing. Display fallbacks do not silently enter calculations.
+4. Do not implement logic in both a root compatibility module and a canonical module.
+5. Do not remove compatibility or uncertain code without reference scanning, old/new
+   differential evidence, browser checks, and manual acceptance.
+6. Run `python tools/verify_baseline.py` for every change and add `--e2e` when runtime or
+   frontend behavior can be affected.

-New endpoints must be added to `api_access.required_role` when they need member or admin
-access. New user-owned tables must include `user_id`, an ownership index, and cross-account
-tests. API payload compatibility is protected by the Python and Playwright suites.
+The authoritative migration constraints and handoff procedure are in
+`../docs/migration/原版保真迁移总纲.md` and
+`../docs/migration/人工维护与本地切换指南.md`.
@@ -2,23 +2,27 @@

 一个面向 A 股盘后复盘的本地 Web 工作台。后端使用 Python 访问 Tushare Pro，前端不依赖构建工具。

+本目录是从原版源码逐项移动、机械拆分并完成差分验证的模块化候选，不是依据规格书重新开发的
+第二套产品。人工验收和正式切换前，`webapp/`根目录仍是唯一正式基线；冻结的`next/`不得用于
+部署或后续开发。目录职责见[ARCHITECTURE.md](ARCHITECTURE.md)。
+
 当前包含集合竞价、涨停池、炸板池、跌停板、昨日涨停、涨停表现、市场天梯、板块轮动、题材库、人气热榜、龙虎榜和个人复盘工作区。交易日快照与同步记录保存在本地 SQLite 数据库 `data/review.db`。

 集合竞价中心采用盘前生命周期：9:15 前显示预告，9:15–9:25 明确等待最终竞价，9:25–9:30 自动读取并重试最终竞价筛选，9:30 后停止更新并冻结为复盘归档。当前 Tushare 只提供 9:25 最终竞价快照，不将其表述为动态虚拟撮合行情。

 第三阶段加入了机构席位、席位别名、个股复权日 K、资金流、自选股、涨停原因修订、个股笔记、每日复盘和历史数据回补。

-股票代码在桌面端悬停后会显示分时与日 K 快速预览，默认优先展示日 K；移动端点击代码后从底部打开预览面板。股票详情以及板块、题材、指数详情均可在日 K 与最新分时之间切换。日 K 复用个股详情缓存；分时使用隔离的东方财富分钟图表源和短时内存缓存，只负责展示，不写入主行情、不参与情绪、选股或问天计算。图表源不可用时界面会明确显示“分时不可用”，不会使用日 K 数据模拟分时走势。
+股票代码在桌面端悬停后会显示分时与日 K 快速预览，默认优先展示日 K；移动端点击代码后从底部打开预览面板。股票详情以及板块、题材、指数详情均可在日 K 与最新分时之间切换。日 K 复用个股详情缓存；分时优先使用 iFinD，东方财富仅作隔离的展示兜底，并使用短时内存缓存。图表数据不写入主行情、不参与情绪、选股或问天计算；不可用时明确显示“分时不可用”，不会用日 K 模拟分时走势。

-智能选股模块包含 45 日全市场因子库、六阶段市场识别、七套内置策略、受控公式 DSL、自然语言策略编译、候选排名和滚动回测。竞价涨幅、竞价成交额、竞价换手率与竞价量比随因子数据一并同步，可用于自定义公式和历史回测。首次使用需在页面点击“同步因子数据”。未配置 LLM 时使用本地策略模板；配置兼容 API 后自动切换为主模型编译，主模型失败时自动使用辅助模型，两者均支持独立连通性测试。
+智能选股包含六阶段盘后候选、29 套精选策略、自定义公式 DSL、自然语言公式编译、候选排名和滚动回测。阶段与精选策略在当日行情更新后由后台确定性计算；自定义选股由用户手动执行，LLM 只负责编译自然语言条件，不参与候选筛选。竞价、估值、财务、资金、人气和席位等字段按已登记的数据可用性进入因子库，缺失时明确显示覆盖问题。

-每次选股结果会自动进入五交易日持续跟踪，展示 T+1 开盘/收盘、T+3、T+5、最大涨幅与最大回撤。提醒中心支持手工日期提醒，并在策略首日反馈和五日跟踪完成时生成账号私有的站内提醒。
+候选只有经用户手动加入后才进入五交易日持续跟踪，展示 T+1 开盘/收盘、T+3、T+5、最大涨幅与最大回撤。提醒中心支持手工日期提醒，并在策略首日反馈和五日跟踪完成时生成账号私有的站内提醒。

 问师模块会读取当前复盘、近十日市场情绪、涨跌停、昨日反馈、板块轮动、市场阶段、龙虎榜和指定个股数据，再按选中的游资思维 Skill 进行单师对话。对话记录按账号、老师和交易日期保存在服务端；主模型不可用时自动切换辅助模型。

 新增公开问师角色时，在 `游资skills` 下增加一个包含 `SKILL.md` 的独立目录，并在 `游资skills/mentor_catalog.json` 中登记素材等级与结构质检。管理员私有角色放在 `data/private-mentor-skills`，该目录不进入 Git 或 Docker 镜像，且只会出现在管理员的问师列表中。系统会从 Skill 的 frontmatter、一级标题、核心模型和引用语中自动生成角色信息，无需修改注册代码。

-问天模块包含三个相互独立的部分：观势以市场数据生成三才六爻，用于观察“势”，行情缺失或自动取象明显偏差时可显式手动校准六爻，人工结果与自动来源严格区分；观气依据干支、精确节气、五运六气及客主加临关系观察“运”，行业五行仅作传统取象归类；观心通过30秒静心、六次三枚铜钱起卦、察念和解卦完成一次不输入问题的问心仪式。卦象、干支、节气与气机关系均由本地确定性程序计算，LLM只负责解释，不参与起卦或改动结果。
+问天模块包含三个相互独立的部分：观势以市场数据生成三才六爻，用于观察“势”，行情缺失或自动取象明显偏差时可显式手动校准六爻，人工结果与自动来源严格区分；观气依据干支、精确节气、五运六气及客主加临关系观察“运”，行业五行仅作传统取象归类；观心先准备1秒，再完成5轮“吸3秒、顿2秒、呼4秒”，随后以六次三枚铜钱起卦、察念和解卦完成一次不输入问题的问心仪式。卦象、干支、节气与气机关系均由本地确定性程序计算，LLM只负责解释，不参与起卦或改动结果。

 问天模块使用项目本地的 `lunar-python` 计算历法，并使用 `data/iching_zh.json` 中的固定六十四卦、卦辞和爻辞。第三方授权见 `THIRD_PARTY_NOTICES.md`。

@@ -27,7 +31,7 @@
 ## 启动

 ```powershell
-cd webapp
+cd webapp\app
 python -m pip install -r requirements.txt
 python server.py
 ```
@@ -1,12 +1,15 @@
 # Governance Registries

-These registries describe the approved product surface during architecture migration.
+These registries describe the approved product surface of the modular preservation candidate.

 - `pages.config.json`: primary page identity, navigation group, access expectation, scrolling,
  and mobile composition policy.
 - `features.config.json`: feature ownership, backend access class, data scope, and availability.
- `api.config.json`: transitional inventory of current routes, generated from `server.py` and
-  assigned to a feature owner.
+- `api.config.json`: current routes generated from the preserved `server.py` API surface, with
+  one feature owner and backend access class per route. Its dispatcher implementation lives in
+  `backend/application.py`.
+- `architecture-inventory.json`: generated inventory of candidate pages, routes, tables,
+  providers, model entry points, CSS layers, and remaining code hotspots.
 - `data-fields.config.json`: canonical data products, provider eligibility, intended use, and
  known blocked datasets.
 - `data-quality.config.json`: freshness, coverage, units, adjustment, point-in-time, and
@@ -14,13 +17,15 @@ These registries describe the approved product surface during architecture migra
 - `jobs.config.json`: background schedules, dependencies, lock keys, retry policy, timeouts,
  and output versions.

-During Stage 04 these files are contract inputs, not runtime replacements. Backend access in
-`api_access.py` remains authoritative until the HTTP governance phase switches it atomically.
-Frontend visibility remains a presentation concern and never grants backend access.
+The registries are governance contracts, not substitutes for runtime authorization. Backend
+access in `backend/http/routes.py` is authoritative; frontend visibility is only a presentation
+concern and never grants access.

 Regenerate the transitional API inventory after a route change:

 ```shell
 python tools/build_api_registry.py
 python tools/build_api_registry.py --check
+python tools/build_architecture_inventory.py
+python tools/build_architecture_inventory.py --check
 ```
@@ -0,0 +1,336 @@
+{
+  "schema_version": 1,
+  "captured_from": "app modular preservation candidate",
+  "runtime": {
+    "http_server": "http.server.ThreadingHTTPServer",
+    "application_processes": 1,
+    "database": "SQLite WAL",
+    "frontend": "build-free HTML/CSS/JavaScript",
+    "container_port": 8765
+  },
+  "counts": {
+    "primary_pages": 16,
+    "api_exact_paths": 53,
+    "api_prefixes": 0,
+    "api_patterns": 11,
+    "database_tables": 36
+  },
+  "pages": [
+    {
+      "id": "sentimentCycleView",
+      "title": "情绪周期"
+    },
+    {
+      "id": "limitPool",
+      "title": "涨停池"
+    },
+    {
+      "id": "brokenView",
+      "title": "炸板池"
+    },
+    {
+      "id": "downView",
+      "title": "跌停板"
+    },
+    {
+      "id": "yesterdayView",
+      "title": "昨日涨停"
+    },
+    {
+      "id": "performanceView",
+      "title": "涨停表现"
+    },
+    {
+      "id": "ladderView",
+      "title": "市场天梯"
+    },
+    {
+      "id": "rotationView",
+      "title": "板块轮动"
+    },
+    {
+      "id": "auctionView",
+      "title": "集合竞价"
+    },
+    {
+      "id": "themeLibraryView",
+      "title": "题材库"
+    },
+    {
+      "id": "popularityView",
+      "title": "人气热榜"
+    },
+    {
+      "id": "dragonView",
+      "title": "龙虎榜"
+    },
+    {
+      "id": "screenerView",
+      "title": "智能选股"
+    },
+    {
+      "id": "mentorView",
+      "title": "问师"
+    },
+    {
+      "id": "heavenView",
+      "title": "问天"
+    },
+    {
+      "id": "reviewWorkspaceView",
+      "title": "我的复盘"
+    }
+  ],
+  "api": {
+    "exact": [
+      "/api/account/birth-profile",
+      "/api/account/password",
+      "/api/account/status",
+      "/api/admin/membership",
+      "/api/admin/refresh",
+      "/api/admin/settings",
+      "/api/admin/settings/test",
+      "/api/alerts",
+      "/api/alerts/read-all",
+      "/api/assistant/chat",
+      "/api/assistant/messages",
+      "/api/auction",
+      "/api/auth/login",
+      "/api/auth/logout",
+      "/api/auth/me",
+      "/api/auth/register",
+      "/api/backfill",
+      "/api/chart/intraday",
+      "/api/dashboard",
+      "/api/dragon-tiger",
+      "/api/dragon-tiger/profiles",
+      "/api/health",
+      "/api/heaven/hexagram",
+      "/api/heaven/interpret",
+      "/api/heaven/personal",
+      "/api/heaven/readings",
+      "/api/heaven/sector-phases",
+      "/api/heaven/setup",
+      "/api/mentors/chat",
+      "/api/mentors/messages",
+      "/api/mentors/preferences",
+      "/api/mentors/setup",
+      "/api/notes",
+      "/api/popularity",
+      "/api/realtime-aggregate/health",
+      "/api/reasons",
+      "/api/rotation/history",
+      "/api/rotation/members",
+      "/api/screener/compile",
+      "/api/screener/run",
+      "/api/screener/setup",
+      "/api/screener/strategies",
+      "/api/screener/sync",
+      "/api/screener/tracking",
+      "/api/screener/tracking/refresh",
+      "/api/search",
+      "/api/search/detail",
+      "/api/seat-aliases",
+      "/api/sentiment/history",
+      "/api/themes",
+      "/api/themes/detail",
+      "/api/trades",
+      "/api/watchlist"
+    ],
+    "prefixes": [],
+    "patterns": [
+      "/api/alerts/(\\d+)",
+      "/api/alerts/(\\d+)/read",
+      "/api/heaven/readings/(\\d+)",
+      "/api/heaven/sector-phases/(.+)",
+      "/api/notes/(\\d+)",
+      "/api/screener/strategies/(\\d+)",
+      "/api/screener/tracking/(\\d+)",
+      "/api/stock/(\\d{6})",
+      "/api/stock/(\\d{6})/preview",
+      "/api/trades/(\\d+)",
+      "/api/watchlist/(\\d{6})"
+    ]
+  },
+  "database_tables": [
+    "users",
+    "user_sessions",
+    "user_credentials",
+    "user_birth_profiles",
+    "system_settings",
+    "llm_usage",
+    "dashboard_snapshots",
+    "sync_runs",
+    "data_snapshots",
+    "watchlist",
+    "review_notes",
+    "reason_overrides",
+    "seat_aliases",
+    "sector_phase_overrides",
+    "stock_master",
+    "daily_bars",
+    "benchmark_bars",
+    "daily_indicators",
+    "fundamental_indicators",
+    "moneyflow_daily",
+    "auction_factors",
+    "earnings_events",
+    "popularity_factors",
+    "lhb_institution_daily",
+    "screener_strategies",
+    "screener_runs",
+    "mentor_messages",
+    "mentor_preferences",
+    "wencai_saved_queries",
+    "strategy_tracks",
+    "alerts",
+    "trade_entries",
+    "assistant_messages",
+    "heaven_readings",
+    "job_runs",
+    "schema_migrations"
+  ],
+  "background_job_methods": [
+    "_background_refresh_tick"
+  ],
+  "external_data_adapters": [
+    {
+      "provider": "tushare",
+      "path": "backend/data/providers/tushare_client.py",
+      "runtime_role": "primary deterministic market data"
+    },
+    {
+      "provider": "ifind",
+      "path": "backend/data/providers/ifind_client.py",
+      "runtime_role": "realtime, charts, snapshots, enrichment"
+    },
+    {
+      "provider": "eastmoney",
+      "path": "backend/features/market/charts.py",
+      "runtime_role": "display chart fallback"
+    },
+    {
+      "provider": "eastmoney",
+      "path": "backend/data/realtime.py",
+      "runtime_role": "isolated realtime observation"
+    },
+    {
+      "provider": "tencent",
+      "path": "backend/data/realtime.py",
+      "runtime_role": "index observation fallback"
+    }
+  ],
+  "llm_entrypoints": [
+    {
+      "function": "stream_with_mentor",
+      "path": "backend/features/mentor/agent.py"
+    },
+    {
+      "function": "interpret_heaven",
+      "path": "backend/features/heaven/agent.py"
+    },
+    {
+      "function": "stream_review_assistant",
+      "path": "backend/features/review/agent.py"
+    },
+    {
+      "function": "compile_strategy_with_llm",
+      "path": "backend/features/screener/compiler.py"
+    },
+    {
+      "function": "test_llm_connection",
+      "path": "backend/features/screener/compiler.py"
+    }
+  ],
+  "css_layers": [
+    "/shared/tokens.css?v=20260729-1",
+    "/styles/styles.css",
+    "/styles/renovation.css?v=20260725-5",
+    "/styles/redesign-v2.css?v=20260728-1",
+    "/styles/design-system.css?v=20260728-4",
+    "/styles/theme.css?v=20260728-2",
+    "/pages/heaven/page.css?v=20260728-7"
+  ],
+  "code_hotspots": [
+    {
+      "path": "frontend/styles/styles.css",
+      "bytes": 361780,
+      "lines": 15465
+    },
+    {
+      "path": "frontend/styles/redesign-v2.css",
+      "bytes": 263539,
+      "lines": 8570
+    },
+    {
+      "path": "frontend/index.html",
+      "bytes": 135019,
+      "lines": 1892
+    },
+    {
+      "path": "backend/features/screener/engine.py",
+      "bytes": 108552,
+      "lines": 2213
+    },
+    {
+      "path": "backend/data/providers/tushare_client.py",
+      "bytes": 94329,
+      "lines": 2175
+    },
+    {
+      "path": "frontend/app.js",
+      "bytes": 91151,
+      "lines": 1939
+    },
+    {
+      "path": "frontend/pages/heaven/page.js",
+      "bytes": 88322,
+      "lines": 1830
+    },
+    {
+      "path": "frontend/styles/renovation.css",
+      "bytes": 83949,
+      "lines": 1553
+    },
+    {
+      "path": "frontend/pages/heaven/page.css",
+      "bytes": 73222,
+      "lines": 1084
+    },
+    {
+      "path": "backend/features/heaven/service.py",
+      "bytes": 64421,
+      "lines": 1303
+    },
+    {
+      "path": "backend/features/market/insights.py",
+      "bytes": 58150,
+      "lines": 1314
+    },
+    {
+      "path": "frontend/pages/market/runtime.js",
+      "bytes": 57053,
+      "lines": 1333
+    },
+    {
+      "path": "backend/features/heaven/engine.py",
+      "bytes": 51670,
+      "lines": 1181
+    },
+    {
+      "path": "backend/application.py",
+      "bytes": 50934,
+      "lines": 1165
+    },
+    {
+      "path": "frontend/styles/theme.css",
+      "bytes": 36427,
+      "lines": 1253
+    },
+    {
+      "path": "database.py",
+      "bytes": 34013,
+      "lines": 746
+    }
+  ]
+}
@@ -0,0 +1,68 @@
+from __future__ import annotations
+
+import json
+import subprocess
+import sys
+import unittest
+from pathlib import Path
+
+from tools.build_api_registry import build as build_api_registry
+from tools.build_architecture_inventory import build as build_architecture_inventory
+
+
+ROOT = Path(__file__).resolve().parents[1]
+
+
+class MaintenanceToolTests(unittest.TestCase):
+    def test_generated_candidate_registries_are_current(self) -> None:
+        expected_api = json.loads(
+            (ROOT / "config" / "api.config.json").read_text(encoding="utf-8")
+        )
+        expected_architecture = json.loads(
+            (ROOT / "config" / "architecture-inventory.json").read_text(
+                encoding="utf-8"
+            )
+        )
+        self.assertEqual(expected_api, build_api_registry())
+        self.assertEqual(expected_architecture, build_architecture_inventory())
+
+    def test_baseline_verifier_uses_the_candidate_frontend(self) -> None:
+        source = (ROOT / "tools" / "verify_baseline.py").read_text(encoding="utf-8")
+        self.assertIn('FRONTEND_ROOT = ROOT / "frontend"', source)
+        self.assertIn('FRONTEND_ROOT.rglob("*.js")', source)
+        self.assertIn("start_e2e_server()", source)
+        self.assertIn("stop_e2e_server(server)", source)
+        self.assertNotIn('"static/app.js"', source)
+
+    def test_every_tool_has_a_non_mutating_help_path(self) -> None:
+        for path in sorted((ROOT / "tools").glob("*.py")):
+            if path.name.startswith("_"):
+                continue
+            with self.subTest(tool=path.name):
+                result = subprocess.run(
+                    [sys.executable, str(path), "--help"],
+                    cwd=ROOT,
+                    capture_output=True,
+                    text=True,
+                    timeout=15,
+                    check=False,
+                )
+                self.assertEqual(result.returncode, 0, result.stderr)
+
+    def test_migration_only_tools_are_explicitly_classified(self) -> None:
+        readme = (ROOT / "tools" / "README.md").read_text(encoding="utf-8")
+        for name in (
+            "build_preservation_manifest.py",
+            "move_class_methods.py",
+            "split_frontend_runtime.py",
+        ):
+            self.assertIn(name, readme)
+        manifest = (ROOT / "tools" / "build_preservation_manifest.py").read_text(
+            encoding="utf-8"
+        )
+        self.assertNotIn('TARGET = ROOT / "app"', manifest)
+        self.assertIn('required=True', manifest)
+
+
+if __name__ == "__main__":
+    unittest.main()
@@ -0,0 +1,36 @@
+# Candidate tools
+
+Run these commands from `webapp/app/`. The tools are separated by responsibility so a
+maintenance command cannot be mistaken for a historical migration rewrite.
+
+## Normal maintenance
+
+- `python tools/verify_baseline.py`: candidate unit tests, registry checks, every frontend
+  JavaScript syntax check, Git whitespace check, and read-only SQLite integrity check.
+- `python tools/verify_baseline.py --e2e`: the same checks plus Playwright. The verifier owns
+  the local static-server lifecycle so the command exits cleanly on Windows.
+- `python tools/build_api_registry.py [--check]`: generate or verify
+  `config/api.config.json` from `backend/application.py`.
+- `python tools/build_architecture_inventory.py [--check]`: generate or verify
+  `config/architecture-inventory.json` from the candidate source tree.
+
+## Acceptance and differential checks
+
+- `run_preservation_runtime.py`: start an isolated original or candidate runtime with an
+  explicitly selected data directory and port.
+- `compare_preservation_apis.py`: compare authenticated responses from two isolated runtimes.
+- `compare_preservation_databases.py`: compare schema and selected table contents from two
+  SQLite copies.
+
+These tools require explicit paths and do not select the production database automatically.
+
+## Migration-only tools
+
+- `build_preservation_manifest.py`: builds an exact-copy manifest for a specified source and
+  target. `--output` is mandatory so committed historical evidence is not overwritten.
+- `move_class_methods.py`: mechanically moves named class methods between explicit files.
+- `split_frontend_runtime.py`: reproduces the one-time Slice 10 split. It refuses to write
+  unless `--apply` is supplied and is not a normal maintenance command.
+
+The migration-only tools are retained for audit and reproducibility. They are not part of
+application startup, normal testing, or future feature development.
@@ -9,7 +9,7 @@ from typing import Any


 ROOT = Path(__file__).resolve().parents[1]
-OUTPUT = ROOT / "docs" / "governance" / "architecture-inventory.json"
+OUTPUT = ROOT / "config" / "architecture-inventory.json"


 def relative(path: Path) -> str:
@@ -51,8 +51,15 @@ def api_inventory(server: str) -> dict[str, list[str]]:
    return {"exact": exact, "prefixes": prefixes, "patterns": patterns}


-def database_inventory(database: str) -> list[str]:
-    return re.findall(r"CREATE TABLE IF NOT EXISTS\s+([a-zA-Z0-9_]+)", database)
+def database_inventory(sources: list[str]) -> list[str]:
+    tables: list[str] = []
+    for database in sources:
+        tables.extend(
+            re.findall(
+                r"CREATE TABLE IF NOT EXISTS\s+([a-zA-Z0-9_]+)", database
+            )
+        )
+    return list(dict.fromkeys(tables))


 def python_functions(path: str, prefixes: tuple[str, ...]) -> list[str]:
@@ -70,11 +77,13 @@ def css_layers(html: str) -> list[str]:

 def code_hotspots() -> list[dict[str, Any]]:
    candidates = [
-        "server.py",
+        "backend/application.py",
        "database.py",
-        "screener.py",
-        "market_insights.py",
-        "tushare_client.py",
+        "backend/features/screener/engine.py",
+        "backend/features/market/insights.py",
+        "backend/data/providers/tushare_client.py",
+        "backend/features/heaven/service.py",
+        "backend/features/heaven/engine.py",
        "frontend/index.html",
        "frontend/app.js",
        "frontend/styles/styles.css",
@@ -88,6 +97,8 @@ def code_hotspots() -> list[dict[str, Any]]:
    rows = []
    for name in candidates:
        path = ROOT / name
+        if not path.is_file():
+            continue
        rows.append(
            {
                "path": name,
@@ -100,14 +111,21 @@ def code_hotspots() -> list[dict[str, Any]]:

 def build() -> dict[str, Any]:
    html = source("frontend/index.html")
-    server = source("server.py")
-    database = source("database.py")
+    server = source("backend/application.py")
+    database_sources = [source("database.py")]
+    database_sources.extend(
+        path.read_text(encoding="utf-8")
+        for path in sorted((ROOT / "backend" / "database" / "migrations").glob("m*.py"))
+    )
+    database_sources.append(
+        source("backend/database/migrations/runner.py")
+    )
    pages = page_inventory(html)
    api = api_inventory(server)
-    tables = database_inventory(database)
+    tables = database_inventory(database_sources)
    return {
        "schema_version": 1,
-        "captured_from": "governed source tree",
+        "captured_from": "app modular preservation candidate",
        "runtime": {
            "http_server": "http.server.ThreadingHTTPServer",
            "application_processes": 1,
@@ -126,21 +144,22 @@ def build() -> dict[str, Any]:
        "api": api,
        "database_tables": tables,
        "background_job_methods": python_functions(
-            "server.py", ("_background", "_run_background", "_schedule_", "run_automatic")
+            "backend/application.py",
+            ("_background", "_run_background", "_schedule_", "run_automatic"),
        ),
        "external_data_adapters": [
-            {"provider": "tushare", "path": "tushare_client.py", "runtime_role": "primary deterministic market data"},
-            {"provider": "ifind", "path": "ifind_client.py", "runtime_role": "realtime, charts, snapshots, enrichment"},
-            {"provider": "eastmoney", "path": "chart_data_provider.py", "runtime_role": "display chart fallback"},
-            {"provider": "eastmoney", "path": "realtime_aggregator.py", "runtime_role": "isolated realtime observation"},
-            {"provider": "tencent", "path": "realtime_aggregator.py", "runtime_role": "index observation fallback"},
+            {"provider": "tushare", "path": "backend/data/providers/tushare_client.py", "runtime_role": "primary deterministic market data"},
+            {"provider": "ifind", "path": "backend/data/providers/ifind_client.py", "runtime_role": "realtime, charts, snapshots, enrichment"},
+            {"provider": "eastmoney", "path": "backend/features/market/charts.py", "runtime_role": "display chart fallback"},
+            {"provider": "eastmoney", "path": "backend/data/realtime.py", "runtime_role": "isolated realtime observation"},
+            {"provider": "tencent", "path": "backend/data/realtime.py", "runtime_role": "index observation fallback"},
        ],
        "llm_entrypoints": [
-            {"function": "stream_with_mentor", "path": "mentor_agent.py"},
-            {"function": "interpret_heaven", "path": "heaven_agent.py"},
-            {"function": "stream_review_assistant", "path": "assistant_agent.py"},
-            {"function": "compile_strategy_with_llm", "path": "llm_strategy.py"},
-            {"function": "test_llm_connection", "path": "llm_strategy.py"},
+            {"function": "stream_with_mentor", "path": "backend/features/mentor/agent.py"},
+            {"function": "interpret_heaven", "path": "backend/features/heaven/agent.py"},
+            {"function": "stream_review_assistant", "path": "backend/features/review/agent.py"},
+            {"function": "compile_strategy_with_llm", "path": "backend/features/screener/compiler.py"},
+            {"function": "test_llm_connection", "path": "backend/features/screener/compiler.py"},
        ],
        "css_layers": css_layers(html),
        "code_hotspots": code_hotspots(),
@@ -154,7 +173,10 @@ def main() -> int:
    rendered = json.dumps(build(), ensure_ascii=False, indent=2) + "\n"
    if args.check:
        if not OUTPUT.exists() or OUTPUT.read_text(encoding="utf-8") != rendered:
-            raise SystemExit("architecture inventory is stale; run tools/build_architecture_inventory.py")
+            raise SystemExit(
+                "architecture inventory is stale; run "
+                "tools/build_architecture_inventory.py"
+            )
        print("Architecture inventory is current.")
        return 0
    OUTPUT.parent.mkdir(parents=True, exist_ok=True)
@@ -1,14 +1,13 @@
 from __future__ import annotations

+import argparse
 import hashlib
 import json
 from datetime import datetime
 from pathlib import Path


-ROOT = Path(__file__).resolve().parents[1]
-TARGET = ROOT / "app"
-OUTPUT = ROOT / "docs" / "migration" / "原版资产清单.json"
+REPOSITORY_ROOT = Path(__file__).resolve().parents[2]

 DIRECTORIES = (
    "backend",
@@ -69,33 +68,57 @@ def digest(path: Path) -> str:
    return checksum.hexdigest()


-def source_files() -> list[Path]:
-    files = [ROOT / name for name in ROOT_FILES]
+def display_path(path: Path, base: Path) -> str:
+    try:
+        return path.relative_to(base).as_posix() or "."
+    except ValueError:
+        return path.as_posix()
+
+
+def source_files(source_root: Path) -> list[Path]:
+    files = [source_root / name for name in ROOT_FILES]
    for directory in DIRECTORIES:
+        root = source_root / directory
+        if not root.is_dir():
+            continue
        files.extend(
            path
-            for path in (ROOT / directory).rglob("*")
+            for path in root.rglob("*")
            if path.is_file() and "__pycache__" not in path.parts
        )
-    return sorted(set(files), key=lambda path: path.relative_to(ROOT).as_posix())
+    return sorted(
+        set(files), key=lambda path: path.relative_to(source_root).as_posix()
+    )


-def build_manifest() -> dict[str, object]:
+def build_manifest(
+    source_root: Path,
+    target_root: Path,
+    source_commit: str,
+) -> dict[str, object]:
    assets = []
    mismatches = []
-    for source in source_files():
-        relative = source.relative_to(ROOT)
-        target = TARGET / relative
-        source_hash = digest(source)
+    for source in source_files(source_root):
+        relative = source.relative_to(source_root)
+        target = target_root / relative
+        source_exists = source.is_file()
+        source_hash = digest(source) if source_exists else ""
        target_hash = digest(target) if target.is_file() else ""
-        status = "identical" if source_hash == target_hash else "mismatch"
+        if not source_exists:
+            status = "missing_source"
+        elif not target.is_file():
+            status = "missing_target"
+        elif source_hash == target_hash:
+            status = "identical"
+        else:
+            status = "mismatch"
        if status != "identical":
            mismatches.append(relative.as_posix())
        assets.append(
            {
                "source": relative.as_posix(),
-                "target": f"app/{relative.as_posix()}",
-                "bytes": source.stat().st_size,
+                "target": display_path(target, REPOSITORY_ROOT),
+                "bytes": source.stat().st_size if source_exists else 0,
                "sha256": source_hash,
                "disposition": "original_copy_pending_move",
                "status": status,
@@ -104,9 +127,9 @@ def build_manifest() -> dict[str, object]:
    return {
        "schema_version": 1,
        "generated_at": datetime.now().astimezone().isoformat(timespec="seconds"),
-        "source_commit": "41329943c4878fc09ed82ec376eb93ab151e4092",
-        "source_root": ".",
-        "target_root": "app",
+        "source_commit": source_commit,
+        "source_root": display_path(source_root, REPOSITORY_ROOT),
+        "target_root": display_path(target_root, REPOSITORY_ROOT),
        "excluded": [
            "next/",
            "data/review.db and SQLite sidecars",
@@ -123,14 +146,34 @@ def build_manifest() -> dict[str, object]:


 def main() -> int:
-    manifest = build_manifest()
-    OUTPUT.write_text(
+    parser = argparse.ArgumentParser(
+        description=(
+            "Build an exact-copy manifest for a preservation migration stage. "
+            "This is a migration-only tool; it does not describe the final moved layout."
+        )
+    )
+    parser.add_argument("--source-root", type=Path, default=REPOSITORY_ROOT)
+    parser.add_argument("--target-root", type=Path, default=REPOSITORY_ROOT / "app")
+    parser.add_argument("--source-commit", default="")
+    parser.add_argument(
+        "--output",
+        type=Path,
+        required=True,
+        help="write to a new audit path; do not overwrite committed slice evidence",
+    )
+    args = parser.parse_args()
+    source_root = args.source_root.resolve()
+    target_root = args.target_root.resolve()
+    output = args.output.resolve()
+    manifest = build_manifest(source_root, target_root, args.source_commit)
+    output.parent.mkdir(parents=True, exist_ok=True)
+    output.write_text(
        json.dumps(manifest, ensure_ascii=False, indent=2) + "\n",
        encoding="utf-8",
    )
    print(
        f"assets={manifest['asset_count']} mismatches={manifest['mismatch_count']} "
-        f"output={OUTPUT}"
+        f"output={output}"
    )
    return 1 if manifest["mismatch_count"] else 0

@@ -1,5 +1,6 @@
 from __future__ import annotations

+import argparse
 import hashlib
 import json
 import re
@@ -91,6 +92,20 @@ def extract_written_chunks(paths: list[Path]) -> dict[tuple[int, int], str]:


 def main() -> None:
+    parser = argparse.ArgumentParser(
+        description="Reproduce the one-time Slice 10 frontend source split"
+    )
+    parser.add_argument(
+        "--apply",
+        action="store_true",
+        help="perform the historical split; this rewrites frontend runtime files",
+    )
+    args = parser.parse_args()
+    if not args.apply:
+        parser.error(
+            "this migration-only tool rewrites files; pass --apply only when "
+            "reproducing Slice 10 from its documented pre-split checkpoint"
+        )
    source_path = ORIGINAL_STATIC / "app.js"
    target_path = FRONTEND_ROOT / "app.js"
    source_bytes = source_path.read_bytes()
@@ -5,10 +5,15 @@ import shutil
 import sqlite3
 import subprocess
 import sys
+import time
+import urllib.error
+import urllib.request
 from pathlib import Path


 ROOT = Path(__file__).resolve().parents[1]
+FRONTEND_ROOT = ROOT / "frontend"
+E2E_URL = "http://127.0.0.1:8876/index.html"


 def run(label: str, command: list[str]) -> None:
@@ -28,8 +33,63 @@ def verify_database() -> None:
    print(f"\n[database] integrity_check=ok size={database.stat().st_size}")


+def url_reachable(url: str) -> bool:
+    try:
+        with urllib.request.urlopen(url, timeout=1) as response:
+            return response.status == 200
+    except (OSError, urllib.error.URLError):
+        return False
+
+
+def start_e2e_server() -> subprocess.Popen[bytes] | None:
+    if url_reachable(E2E_URL):
+        print(f"\n[e2e-server] reusing {E2E_URL}")
+        return None
+    creation_flags = subprocess.CREATE_NO_WINDOW if sys.platform == "win32" else 0
+    process = subprocess.Popen(
+        [
+            sys.executable,
+            "-m",
+            "http.server",
+            "8876",
+            "--bind",
+            "127.0.0.1",
+            "--directory",
+            "frontend",
+        ],
+        cwd=ROOT,
+        stdout=subprocess.DEVNULL,
+        stderr=subprocess.DEVNULL,
+        creationflags=creation_flags,
+    )
+    deadline = time.monotonic() + 15
+    while time.monotonic() < deadline:
+        if process.poll() is not None:
+            raise RuntimeError("Playwright static server exited before becoming ready")
+        if url_reachable(E2E_URL):
+            print(f"\n[e2e-server] started {E2E_URL}")
+            return process
+        time.sleep(0.1)
+    stop_e2e_server(process)
+    raise RuntimeError("Playwright static server did not become ready within 15 seconds")
+
+
+def stop_e2e_server(process: subprocess.Popen[bytes] | None) -> None:
+    if process is None:
+        return
+    process.terminate()
+    try:
+        process.wait(timeout=5)
+    except subprocess.TimeoutExpired:
+        process.kill()
+        process.wait(timeout=5)
+    print("\n[e2e-server] stopped")
+
+
 def main() -> int:
-    parser = argparse.ArgumentParser(description="Verify the governance regression baseline")
+    parser = argparse.ArgumentParser(
+        description="Verify the modular preservation candidate"
+    )
    parser.add_argument(
        "--e2e",
        action="store_true",
@@ -38,19 +98,37 @@ def main() -> int:
    args = parser.parse_args()

    run("python", [sys.executable, "-m", "unittest", "discover", "-s", "tests"])
+    run(
+        "api-registry",
+        [sys.executable, "tools/build_api_registry.py", "--check"],
+    )
+    run(
+        "architecture-inventory",
+        [sys.executable, "tools/build_architecture_inventory.py", "--check"],
+    )
    node = shutil.which("node")
    if not node:
        raise RuntimeError("node is required for JavaScript syntax checks")
-    for script in ("static/app.js", "static/heaven-loading-v2.js"):
-        run("javascript", [node, "--check", script])
+    scripts = sorted(FRONTEND_ROOT.rglob("*.js"))
+    if not scripts:
+        raise RuntimeError(f"no JavaScript files found under {FRONTEND_ROOT}")
+    for script in scripts:
+        run(
+            "javascript",
+            [node, "--check", script.relative_to(ROOT).as_posix()],
+        )
    run("patch", ["git", "diff", "--check"])
    verify_database()

    if args.e2e:
-        npm = shutil.which("npm.cmd" if sys.platform == "win32" else "npm")
-        if not npm:
-            raise RuntimeError("npm is required for the Playwright suite")
-        run("playwright", [npm, "run", "test:e2e"])
+        npx = shutil.which("npx.cmd" if sys.platform == "win32" else "npx")
+        if not npx:
+            raise RuntimeError("npx is required for the Playwright suite")
+        server = start_e2e_server()
+        try:
+            run("playwright", [npx, "playwright", "test", "--reporter=dot"])
+        finally:
+            stop_e2e_server(server)
    print("\nBaseline verification passed.")
    return 0

@@ -2,6 +2,7 @@

 > 基线：`dec3cd12365df5e7d4c391369f14c457cf56d7d9`
 > 候选回档标签：`xiaobai-preservation-slice-11-candidate-20260731`
+> 严格审计候选标签：`xiaobai-preservation-slice-11-audit-candidate-20260731`
 > 当前结论：自动验收完成，等待用户人工验收；尚未执行正式切换、Docker或NAS部署

 ## 1. 本切片做了什么
@@ -30,7 +31,7 @@
 | 验证 | 结果 |
 |---|---:|
 | 原版`python -m unittest discover -s tests -q` | 231项通过 |
-| 迁移版`python -m unittest discover -s tests -q` | 298项通过 |
+| 迁移版`python -m unittest discover -s tests -q` | 302项通过 |
 | 历史切片与前端/清理专项复核 | 通过 |
 | 迁移版JavaScript语法检查 | 24个文件通过 |
 | `npx.cmd playwright test --reporter=dot` | 45项通过（1.6分钟） |
@@ -44,6 +45,12 @@

 完整API和数据库结果分别见`api-diff.json`与`database-diff.json`，二者`all_equal`均为`true`。

+严格完成审计另外验证了候选维护工具本身：API注册表和候选架构清单均可生成并执行
+`--check`；24个前端脚本全部由统一命令发现；迁移期写入工具必须显式指定输出或`--apply`。
+Windows下Playwright托管Python静态服务器会在45项执行后不退出，统一验证器现改为显式启动、
+探活和停止8876；最终同一命令正常退出并明确报告`45 passed`。四项维护工具契约测试使迁移版
+总数从298增加到302。逐项目标矩阵见`completion-audit.md`。
+
 ## 3. 真实浏览器验收

 - 桌面端复核情绪周期、智能选股三个工作区、问天日间/夜间和加载资源；控制台无新增
@@ -67,7 +74,7 @@

 ## 5. 回档与最终确认

-切片11提交和标签只代表“候选可验收”，不代表删除已被永久确认。人工验收前可按
+切片11提交和两个候选标签只代表“候选可验收”，不代表删除已被永久确认。人工验收前可按
 `uncertain-code-audit.md`从切片10标签单独恢复任一候选；不需要回退其他已通过迁移切片。

 用户人工确认后才允许：
@@ -18,8 +18,8 @@
      "endpoint": "/api/account/status",
      "original_status": 200,
      "migrated_status": 200,
-      "original_sha256": "094530da01cbc633de4e2a09d1b3a58689ed9743d2ab2207c495e6c44781fc1e",
-      "migrated_sha256": "094530da01cbc633de4e2a09d1b3a58689ed9743d2ab2207c495e6c44781fc1e",
+      "original_sha256": "0ac1b88affa62671f14ba9ea0bd42155952fdc2e03dba3573c232bb452733bc6",
+      "migrated_sha256": "0ac1b88affa62671f14ba9ea0bd42155952fdc2e03dba3573c232bb452733bc6",
      "equal": true,
      "first_difference": null
    },
@@ -62,8 +62,8 @@
      "endpoint": "/api/auction?trade_date=2026-07-30",
      "original_status": 200,
      "migrated_status": 200,
-      "original_sha256": "5a6b643ab1d7e2285ae5667bca12e6ed887ba4aa7512ba989e0dcfb2032e755b",
-      "migrated_sha256": "5a6b643ab1d7e2285ae5667bca12e6ed887ba4aa7512ba989e0dcfb2032e755b",
+      "original_sha256": "bd75b100485c0afa0cdddc73d1880160eea8c0b8c91f556245c665d591285ed6",
+      "migrated_sha256": "bd75b100485c0afa0cdddc73d1880160eea8c0b8c91f556245c665d591285ed6",
      "equal": true,
      "first_difference": null
    },
@@ -95,8 +95,8 @@
      "endpoint": "/api/dragon-tiger?trade_date=2026-07-30",
      "original_status": 200,
      "migrated_status": 200,
-      "original_sha256": "9d19029f655813ed40a25deb0b3250423a40ae0ee779638ec90f85333f859c6f",
-      "migrated_sha256": "9d19029f655813ed40a25deb0b3250423a40ae0ee779638ec90f85333f859c6f",
+      "original_sha256": "1d6936e33aea2bb4258d1d4d007385626d0c638fc290b66271f00649dba59358",
+      "migrated_sha256": "1d6936e33aea2bb4258d1d4d007385626d0c638fc290b66271f00649dba59358",
      "equal": true,
      "first_difference": null
    },
@@ -216,8 +216,8 @@
      "endpoint": "/api/alerts?status=all&as_of=2026-07-30",
      "original_status": 200,
      "migrated_status": 200,
-      "original_sha256": "01e8f169fbc8700313dbdf8379e682b78d7b372cfbee556e1d68f46cd6cbeecf",
-      "migrated_sha256": "01e8f169fbc8700313dbdf8379e682b78d7b372cfbee556e1d68f46cd6cbeecf",
+      "original_sha256": "b611e936ada5d547c728e33e811630282d1eecd7c1ab1517e22b34902eda032a",
+      "migrated_sha256": "b611e936ada5d547c728e33e811630282d1eecd7c1ab1517e22b34902eda032a",
      "equal": true,
      "first_difference": null
    },
@@ -0,0 +1,72 @@
+# 保真迁移完成度逐项审计
+
+> 审计对象：`webapp/app/`
+> 审计基线：`xiaobai-preservation-slice-11-candidate-20260731`
+> 结论口径：自动证据通过不替代用户人工验收，也不授权部署切换
+
+本审计回答的不是“测试是否为绿色”，而是迁移总纲、批准目录、产品表面和人工维护目标是否
+分别有可复验的证据。状态只使用：`自动闭环`、`人工待验`、`明确保留`和`未执行`。
+
+## 1. 目标与证据矩阵
+
+| 要求 | 当前实现 | 主要证据 | 状态 |
+|---|---|---|---|
+| 原版是唯一产品/视觉基线 | 所有切片从根目录原版复制或机械移动；规格书不覆盖真实行为 | 总纲；切片00-11源码映射 | 自动闭环 |
+| `next/`完全隔离 | `next/`未进入候选导入、资产、测试或部署路径；切片10后Git差分为0 | `next失败冻结记录.md`；Git路径审计 | 自动闭环 |
+| 批准的模块化单体目录 | `backend/{bootstrap,http,features,data,database,jobs,llm}`、`frontend/{shared,pages,styles,vendor}`、`config/data/tests/tools`均已归位 | `app/config/architecture-inventory.json`；切片01-10 | 自动闭环 |
+| 不制造第二套产品实现 | 根入口是兼容外壳；业务方法从原类机械移动，专项测试逐符号比较AST/哈希 | 切片01-10 preservation tests | 自动闭环 |
+| 16个主工作区及内部能力完整 | 页面注册表有16个主工作区；策略跟踪、搜索、详情、提醒、账户和系统管理保留 | `pages.config.json`；前端边界测试；Playwright | 自动闭环 |
+| API路径、权限、错误语义不变 | 53个精确路径和11个正则路径有唯一所有者；鉴权在后端执行 | `api.config.json`；HTTP/权限测试；21端点差分 | 自动闭环 |
+| 数据源有统一边界和质量规则 | 18个数据集登记来源、用途、单位、新鲜度、覆盖率、复权和失败关闭规则 | data registries；DataGateway/quality tests | 自动闭环 |
+| 数据库无损、用户隔离不变 | 36张表由初始schema和有序migration管理；62个schema对象、21张关键表差分相同 | 数据库差分；migration/repository/account-boundary tests | 自动闭环 |
+| 后台任务可追踪和重试 | 3个任务登记调度、锁、超时、重试和输出版本；运行状态持久化 | `jobs.config.json`；job runner tests | 自动闭环 |
+| LLM唯一治理边界 | 问师、问天、复盘助手和策略编译经统一网关执行会员、额度、模型回退、流式和审计规则 | 切片07/09；LLM gateway/stream tests | 自动闭环 |
+| 前端唯一请求出口、Shell和页面职责 | 只有`frontend/shared/api.js`调用`fetch`；Shell、状态、弹窗、页面生命周期及页面模块已归位 | frontend boundary tests；切片10源码重组哈希 | 自动闭环 |
+| CSS、主题、动画和移动行为不改写 | 原七层CSS和问天动画按字节/源码移动；令牌层与加载顺序受测试保护 | CSS governance；切片10/11浏览器证据 | 自动闭环 |
+| 不确定代码先记录再试删 | 2个文件和5个无消费者函数仅为候选试删；历史表兼容责任继续保留 | `uncertain-code-audit.md`；cleanup contract | 人工待验 |
+| 可由人工维护者复验 | 候选README/架构说明、工具分类、API/架构生成器、统一验证命令均可从`app/`独立运行 | `app/tools/README.md`；maintenance tool tests | 自动闭环 |
+| 正式数据库与部署不受影响 | 验收只使用隔离SQLite副本和非8765端口 | 各切片README；运行记录 | 自动闭环 |
+| Docker/NAS和正式入口切换 | 用户已明确本轮不做NAS Docker测试；人工验收前禁止切换 | 迁移状态和切换指南 | 未执行 |
+
+## 2. 结构减法的可量化结果
+
+- 根`server.py`从原版约5,851行降为25行稳定入口；实现归入启动、HTTP和产品领域模块。
+- 根`database.py`从原版2,839行降为746行，只保留初始schema、组合入口及有明确历史兼容责任的
+  方法；各领域查询已移入对应Repository。
+- 20个其他根Python兼容模块均为1-15行导出/模块别名，不含第二套实现。
+- 原9,283行前端运行时按原连续源码范围拆入共享层和13个页面目录；试删后仍由源码重组测试
+  保护，不能悄悄增加、丢失或重排原行为。
+- 已确认没有消费者的`demo_data.py`、旧问天加载文件和5个函数进入可单项恢复的试删候选；
+  `wencai_saved_queries`因旧数据库兼容和账号隔离继续保留。
+
+大文件不等于自动可删。筛选引擎、Tushare适配器、问天确定性计算和原CSS仍是单一有效实现；
+在没有更细的同输入差分与人工视觉确认前继续拆分，反而违反“整理原件而非重拍”的迁移约束。
+它们已登记在`config/architecture-inventory.json`的`code_hotspots`，供后续正常维护按领域处理。
+
+## 3. 本次严格审计发现并关闭的缺口
+
+候选版初次自动验收后，三个维护工具仍带有旧目录假设：基线验证器引用不存在的`static/`，
+架构清单写向不存在的候选文档目录，原样清单生成器假设`app/app`且会因试删文件直接异常。
+这些问题不改变产品页面，却会让接管者无法按指南复验，因此不能视为可维护性完成。
+
+关闭方式：
+
+1. 基线验证器改为检查候选全量测试、两个生成注册表、`frontend/`下全部JavaScript、Git差异和
+   测试数据库完整性。
+2. 候选架构清单固定生成到`config/architecture-inventory.json`，从正式候选路径统计16页、
+   64类路由匹配、36张表、供应商/LLM入口、CSS层和热点文件。
+3. 原样资产清单、方法搬运和前端拆分明确标为迁移期工具；可能写文件的操作必须显式传路径或
+   `--apply`，默认帮助路径不改文件。
+4. 新增维护工具契约测试，保证生成文件新鲜、所有工具`--help`可用，旧`static/`路径不会回归。
+
+## 4. 仍需用户完成的最终裁决
+
+自动部分完成后仍不能执行以下动作：
+
+1. 把切片11的试删候选从“待人工确认”改成“确认废弃”。
+2. 声称所有页面的视觉、交互手感和动画已经由用户确认等价。
+3. 建立最终完成标签，或把`app/`切换为本地/容器正式入口。
+4. 删除根目录原版、正式数据库备份或冻结的失败记录。
+
+本地人工验收入口及逐页清单以`人工维护与本地切换指南.md`为准。发现差异时只回退对应候选，
+不使用破坏性Git操作覆盖原版或正式数据。
@@ -56,8 +56,8 @@
      "table": "alerts",
      "original_count": 2,
      "migrated_count": 2,
-      "original_sha256": "a496857f915d8ba0870274455182a2d7f180a931836b7078e9676dc0e7bbe430",
-      "migrated_sha256": "a496857f915d8ba0870274455182a2d7f180a931836b7078e9676dc0e7bbe430",
+      "original_sha256": "3fd3b7ceb0950d543fba9cebdd2b4294fe187b37a6b8d4755a49428661b3ae1c",
+      "migrated_sha256": "3fd3b7ceb0950d543fba9cebdd2b4294fe187b37a6b8d4755a49428661b3ae1c",
      "equal": true,
      "excluded_columns": []
    },
@@ -77,19 +77,13 @@ python -m unittest discover -s tests -q

 ```powershell
 cd webapp\app
-python -m unittest discover -s tests -q
-npx.cmd playwright test --reporter=dot
+python tools\verify_baseline.py
+python tools\verify_baseline.py --e2e
 ```

-JavaScript和Git差异：
-
-```powershell
-cd webapp\app
-$files = Get-ChildItem frontend -Recurse -Filter *.js
-foreach ($file in $files) { node --check $file.FullName }
-cd ..
-git diff --check
-```
+第一条命令已经包含迁移版全量单元测试、API/架构注册表新鲜度、全部前端JavaScript语法、
+Git空白错误和测试数据库只读完整性检查。第二条额外运行Playwright。工具的日常/验收/迁移期
+分类见`app/tools/README.md`；迁移期工具不是正常开发命令。

 Playwright需要能够启动本机无头Edge。若测试停在浏览器启动前且没有`msedge`进程，先检查执行
 环境是否禁止GUI/无头浏览器进程；这不是页面失败，不要通过删除测试或延长产品超时绕过。
@@ -1,6 +1,6 @@
 {
  "schema_version": 1,
-  "updated_at": "2026-07-31T16:10:00+08:00",
+  "updated_at": "2026-07-31T20:57:42+08:00",
  "status": "awaiting_manual_acceptance",
  "migration_mode": "behavior_preserving_source_migration",
  "source_of_truth": "current_original_webapp_runtime_and_source",
@@ -9,10 +9,10 @@
  "failed_roots": [
    "next"
  ],
-  "current_slice": "slice-11-uncertain-code-audit-final-acceptance-handoff",
+  "current_slice": "slice-11-strict-completion-audit-maintenance-handoff",
  "last_completed_slice": "slice-10-frontend-shell-pages-components-css-mobile",
-  "last_automated_slice": "slice-11-uncertain-code-audit-final-acceptance-handoff",
-  "last_checkpoint": "xiaobai-preservation-slice-11-candidate-20260731",
+  "last_automated_slice": "slice-11-strict-completion-audit-maintenance-handoff",
+  "last_checkpoint": "xiaobai-preservation-slice-11-audit-candidate-20260731",
  "next_action": "user_acceptance_on_local_port_8797_then_confirm_or_restore_each_trial_retirement; do_not_switch_docker_or_nas_before_approval",
  "authoritative_documents": [
    "AGENTS.md",
@@ -1,6 +1,6 @@
 # 小白复盘保真迁移账本

-> 当前状态：切片11自动验收完成，等待用户人工验收；正式切换尚未执行
+> 当前状态：切片11严格完成审计通过，等待用户人工验收；正式切换尚未执行

 本账本是上下文恢复和人工审计的连续记录。任何迁移提交必须在同一提交中更新本文件及
 `保真迁移状态.json`。
@@ -31,6 +31,7 @@
 | 2026-07-31 | `xiaobai-preservation-slice-09-20260731` | 复盘、自选、交易日志、提醒与复盘助手原实现归位 | 自动、API、数据库、账户隔离与浏览器差分通过，进入切片10 |
 | 2026-07-31 | `xiaobai-preservation-slice-10-20260731` | 前端Shell、页面、CSS、动画与移动端职责归位 | 源码重组、自动、API、数据库、桌面、夜间、移动端与动画差分通过，进入切片11 |
 | 2026-07-31 | `xiaobai-preservation-slice-11-candidate-20260731` | 不确定代码审计、全量验收与人工交接准备 | 自动、API、数据库和浏览器验收通过；等待人工确认，未切换部署 |
+| 2026-07-31 | `xiaobai-preservation-slice-11-audit-candidate-20260731` | 逐项目标审计并修复候选维护工具旧路径 | 302项测试、24个脚本、45项Playwright、21项API与62项schema差分通过；仍等待人工确认 |

 ## 资产处置登记

@@ -197,12 +198,25 @@
 - 处置：前3类进入可单项回档的候选试删；`wencai_saved_queries`表和方法因旧库兼容与用户隔离继续保留。
 - 保护方式：全部历史前端源码测试统一使用审计白名单比较，只允许登记的原`app.js`行号缺失，其他差异仍会失败。
 - API与数据库：21个全站只读API全部一致；62个schema对象和21张关键表逐行一致。
- 验收：原版231项、迁移版298项Python测试、24个JavaScript语法检查和45项Playwright（1.6分钟）通过；桌面日间/夜间、390×844五个移动入口和问天三页真实检查通过。
+- 验收：原版231项、迁移版302项Python测试、24个JavaScript语法检查和45项Playwright通过；桌面日间/夜间、390×844五个移动入口和问天三页真实检查通过。
 - 部署边界：未改正式数据库、`.env`、`8765`、Docker或NAS；原版仍是正式运行基线。
 - 候选回档：标签`xiaobai-preservation-slice-11-candidate-20260731`；试删最终结论仍等待用户人工确认。
 - 完整证据：`docs/migration/evidence/slice-11/README.md`。
 - 维护交接：`docs/migration/人工维护与本地切换指南.md`。

+严格完成审计补充：
+
+- 发现候选`verify_baseline.py`、架构清单和原样资产清单仍带有旧`static/`、候选`docs/`及
+  `app/app`路径假设；产品运行不受影响，但人工接管命令不可用，因此属于迁移目标缺口。
+- 修复后统一验证命令检查302项Python测试、两个生成注册表、24个前端脚本、Git空白、SQLite
+  完整性和45项Playwright，并在Windows上显式管理8876测试服务器生命周期。
+- 21个固定API使用同一数据库时点重新比较全部一致；数据库62个schema对象和21张关键表一致，
+  仅按既有规则排除内置策略启动校准的`updated_at`。
+- 候选自身的16页、64类路由匹配、36张表、数据/LLM/CSS入口和热点文件登记在
+  `app/config/architecture-inventory.json`；逐项结论见`completion-audit.md`。
+- 严格审计回档标签为`xiaobai-preservation-slice-11-audit-candidate-20260731`。该标签仍不是最终
+  完成标签，未改变试删候选、人工验收和部署切换边界。
+
 ## 决策记录

 | 日期 | 决策 | 原因 |