fix: 终审修复——墙执行侧接租户管道(端到端回归锁)+seed补email组+措辞与死依赖清理

C-1(必须修): 墙的执行侧没接租户管道——runtime.compiled_wall 全库只有
AIService 的 $DEFLECTION_POOL 一处消费,conversation.py 执行侧全走模块级
DEFAULT_WALL(jiancai 出厂),auto 租户会话实际跑的是 jc 墙。修复:
- manager 构造时 self._wall = runtime.compiled_wall(与 AIService 同源);
- 6 处过滤调用(violates_boundary/no_contact_value/asks_contact 的句内
  与 pending 尾巴两路)、CC 墙模板/替换句、no_contact 收尾选择、
  pivot/deflection、兜底句、contact_ask 触发日志判定全部传 wall=self._wall;
- _build_context_reminder/_contact_completion_note/_cc_confirm_note/
  _cc_script/_cc_wall_template/_lang_note_for 加 wall=None 尾参
  (None = 模块级默认墙,旧测试直调语义不变;流式侧 _contact_completion_note
  双胞调用同传 wall,写回与注入不漂移);
- 模块级 DEFAULT_WALL 视图常量(BOUNDARY_TRIGGER_RE/_CC_*/_EMAIL_NOTE_* 等)
  保留——boundary_filter 出厂默认与旧测试锚定,boundary_filter 模块级
  DEFAULT_WALL 照旧,conversation 会话流内一律租户墙;
- 端到端回归锁 tests/test_wall_execution_tenant.py(两租户真 runtime +
  脚本化假 AI): jc 独有 [€£] pattern 句在 jiancai 拦截替换(收尾=jc
  pivots 池)/auto 放行逐字流出;同一邮箱违规句两租户各自精确收到自己
  no_contact["web"](12 hours vs replies fast!);contact_ask 提醒注入
  各自墙文案互不渗漏。已验证模拟旧 DEFAULT_WALL 行为时三测全败。

I-1: 两租户 seed 补 email 组(以 brand_values.json 实际值为准:
auto=notify_email 环境值,jiancai=品牌邮箱——其 notify_email 记录是
notify@example.com 占位): auto {sender_name: XYH, notify_emails:
[18339827479@163.com]}, jiancai {sender_name: HESUPIPE, notify_emails:
[hesu@hesuu.com]};seed 存裸名,From 头渲染时外层再包【】。补测
test_seed_tenants_ship_email_group: env NOTIFY_EMAIL patch 成异值证明
seed 槽胜出。

M-1/M-2: README/CLAUDE.md "platform.db 热加载"改为"文件级 registry_reload
热加载(Phase 2 升级为 platform.db)";requirements.txt 删 alembic
(全库零引用,grep 实证唯一命中即该行)。

M-3/M-4(挂账): CLAUDE.md Phase 2 清理单补两项——/health 分级健康
(Phase 4 部署批次)、PromptProvider.from_dir FileNotFoundError 带路径上下文。

全量回归: 463 passed(459 基线 + 4 新增),根目录与 tests/ 子目录同数。

Co-Authored-By: Claude Code <noreply@anthropic.com>

CLAUDE.md

@@ -37,7 +37,7 @@
 - 上游吸收：input-placeholder 可配置化（品牌配置组，public.py 下发，widget/chat data-placeholder 管道）；7 个测试文件（test_channel_hygiene / test_chat_rest_ip / test_lead_push_cc / test_cc_integration / test_ip_service_region / test_prompt_static / test_placeholder_static，commit d1aed8e）。
 - 三不带：blog.html（static/ 无此文件）、Redis（app+scripts 零命中）、旧版 whatsapp-cc-completion（零命中，jc 版 cc_completion.py 取代）。
 
-**当前能力一览**：多租户引擎（TenantRuntime 隔离边界，engine 不感知 platform）；站点模型（tenant_config.Site，一租户多站点覆盖）；词表/提示词/品牌全配置化（seed_profiles 出厂包 + platform.db 热加载）；Host 路由（host_middleware，未知 Host 404）；WS 防串库（租户化 client key + 会话站点标）；WS 每 IP 连接上限（conn_limiter，租户级 ws_conn_limit 配置）。
+**当前能力一览**：多租户引擎（TenantRuntime 隔离边界，engine 不感知 platform）；站点模型（tenant_config.Site，一租户多站点覆盖）；词表/提示词/品牌全配置化（seed_profiles 出厂包 + 文件级 registry_reload 热加载，Phase 2 升级为 platform.db）；Host 路由（host_middleware，未知 Host 404）；WS 防串库（租户化 client key + 会话站点标）；WS 每 IP 连接上限（conn_limiter，租户级 ws_conn_limit 配置）。
 
 ## Phase 2 清理单（前序审查挂账，集中列此）
 
@@ -51,6 +51,8 @@
 8. **话术含引号/换行的转义**——编辑器批次（`_format_deflection_pool` docstring 有注）。
 9. **产品决策项**：国名纠正 e164 变化时是否补发更正邮件（上游行为更优）；REST/WS 服务端 IP 口径是否统一（出处：上游吸收审查）。
 10. **avatar 从 seed_profiles 迁 uploads 卷**——上传功能落地时（出处：品牌下发审查）。
+11. **/health 分级健康**——Phase 4 部署批次（进程存活之外按依赖分级上报，健康检查可区分"进程活"与"依赖通"；出处：终审审查 M-3）。
+12. **PromptProvider.from_dir 的 FileNotFoundError 带路径上下文**——提示词目录缺失时异常信息含实际路径，定位不用猜（出处：终审审查 M-4）。
 
 ## Phase 2 计划编写注意
 

README.md

@@ -5,7 +5,7 @@
 **现状能力**（Phase 1 收尾时点）：
 
 - 多租户引擎：TenantRuntime 隔离边界，engine 包不感知平台层，可整体拎走独立部署。
-- 站点模型：一租户多站点（品牌/主题/站点级覆盖），词表、提示词、品牌全部配置化（seed_profiles 出厂包 + platform.db 热加载）。
+- 站点模型：一租户多站点（品牌/主题/站点级覆盖），词表、提示词、品牌全部配置化（seed_profiles 出厂包 + 文件级 registry_reload 热加载，Phase 2 升级为 platform.db）。
 - Host 路由：按域名解析租户，未知 Host 404；WS 通道租户化客户端 key 防串库，另有限流与每 IP 连接上限。
 - 引擎行为：cc 区号补全判定树、多语言墙（中/西/葡借道/俄/阿）、dual-contact 分步双要、CC 强制墙、lead 邮件推送。
 

app/engine/services/conversation.py

@@ -25,6 +25,7 @@ from app.engine.services.contact_pipeline import (
     canon, digits_only, latest_stated_contacts, merge_contacts,
     normalize_contacts, resolve_by_recency, scan_user_messages, split_validate,
 )
+from app.engine.wall_schema import CompiledWall
 
 if TYPE_CHECKING:
     from app.platform.registry import TenantRuntime
@@ -71,6 +72,8 @@ def context_seed_map(config_data: dict | None) -> dict:
 # - 原则: 误触发低害(多一条提醒),漏触发高危(红线失守),词表偏宽;上线后漏词案例回流补词
 # Task 4 词表数据化: 三组触发正则的数据源 = DEFAULT_WALL(jiancai 出厂)
 # .triggers 编译产物(seed_profiles/jiancai/wall.json),上方精修说明随数据留存
+# C-1(终审): 本组模块常量降级为"默认墙视图"——供直调/旧测试按名锚定;
+# 运行期触发词查找一律走租户墙(manager 的 self._wall.triggers)
 BOUNDARY_TRIGGER_RE = DEFAULT_WALL.triggers["boundary"]
 
 # 预算路径路由(评审#9): 命中 BOUNDARY 触发词且含预算词 → 走预算 playbook
@@ -97,11 +100,14 @@ LANG_NOTE_ZH = DEFAULT_WALL.reminders["language_follow"]["zh"]
 LANG_NOTE_ES = DEFAULT_WALL.reminders["language_follow"]["es"]
 
 
-def _lang_note_for(msg: str) -> str:
+def _lang_note_for(msg: str, wall: CompiledWall | None = None) -> str:
+    # C-1(终审): wall=None = 模块级默认墙(LANG_NOTE_* 常量即该默认视图,
+    # 旧测试锚定);manager 一律传租户墙
+    w = wall if wall is not None else DEFAULT_WALL
     if _CJK_RE.search(msg or ""):
-        return LANG_NOTE_ZH
+        return w.reminders["language_follow"]["zh"]
     if _ES_FEATURE_RE.search(msg or ""):
-        return LANG_NOTE_ES
+        return w.reminders["language_follow"]["es"]
     return ""
 
 
@@ -157,45 +163,60 @@ def _cc_intl(wa_state: dict | None) -> str:
     return str((wa_state or {}).get("intl") or cc_completion.format_intl(e164))
 
 
-def _cc_script(wa_state: dict | None, lang: str) -> str:
+def _cc_script(wa_state: dict | None, lang: str,
+               wall: CompiledWall | None = None) -> str:
     """CC 等待态本轮逐字话术(Task 4, spec §4;P3 扩 ES): 有 e164 → 回显
     确认句(intl 由 Python 算好填入,GLM 不拼号);no_ip → 追问句
-    (_CC_ASK_* 服务)。reminder 注入/墙替换/测试断言三方同源,统一走本函数。"""
+    (_CC_ASK_* 服务)。reminder 注入/墙替换/测试断言三方同源,统一走本函数。
+    C-1(终审): wall=None = 模块级默认墙(_CC_ECHO_*/_CC_ASK_* 常量即该
+    默认视图);manager 一律传租户墙。"""
+    w = wall if wall is not None else DEFAULT_WALL
     if str((wa_state or {}).get("e164") or ""):
         intl = _cc_intl(wa_state)
-        echo = {"zh": _CC_ECHO_ZH, "es": _CC_ECHO_ES}.get(lang, _CC_ECHO_EN)
-        return echo.format(intl=intl)
-    return {"zh": _CC_ASK_ZH, "es": _CC_ASK_ES}.get(lang, _CC_ASK_EN)
+        echo = w.cc_templates["echo"]
+        return {"zh": echo["zh"], "es": echo["es"]}.get(
+            lang, echo["en"]).format(intl=intl)
+    ask = w.cc_templates["ask"]
+    return {"zh": ask["zh"], "es": ask["es"]}.get(lang, ask["en"])
 
 
-def _cc_wall_template(wa_state: dict | None, lang: str) -> str:
+def _cc_wall_template(wa_state: dict | None, lang: str,
+                      wall: CompiledWall | None = None) -> str:
     """CC 触发轮墙整段替换句: 回显轮 = ECHO 模板(自带 "Got it —" 衔接头);
-    no_ip 轮 = _CC_WALL_*(ASK 追问句 + 衔接头,沿用 Task 3 形态;P3 扩 ES)。"""
+    no_ip 轮 = _CC_WALL_*(ASK 追问句 + 衔接头,沿用 Task 3 形态;P3 扩 ES)。
+    C-1(终审): wall=None = 模块级默认墙(_CC_WALL_* 常量即该默认视图);
+    manager 一律传租户墙。"""
     if str((wa_state or {}).get("e164") or ""):
-        return _cc_script(wa_state, lang)
-    return {"zh": _CC_WALL_ZH, "es": _CC_WALL_ES}.get(lang, _CC_WALL_EN)
+        return _cc_script(wa_state, lang, wall=wall)
+    w = wall if wall is not None else DEFAULT_WALL
+    tmpl = w.cc_templates["wall"]
+    return {"zh": tmpl["zh"], "es": tmpl["es"]}.get(lang, tmpl["en"])
 
 
-def _cc_confirm_note(wa_state: dict | None, lang: str) -> str:
+def _cc_confirm_note(wa_state: dict | None, lang: str,
+                     wall: CompiledWall | None = None) -> str:
     """cc 等待态确认句注入(Task 4, spec §4/§4b): cc_due 才注入,与邮箱轻补
     cc_settled 串行互斥(两个"逐字使用这句"不同轮竞争)。外壳仿轻补指引:
     ⚠️ 系统指令声明 + 逐字话术 + 客户答复后致谢不追问。回显分支带零号码
     唯一例外(4b-3 让位语义重建: 仅系统提供的补全号原文可写出)。
-    Task 4 词表数据化: 指引话术数据源 = DEFAULT_WALL.reminders.cc_guide
-    ({intl}/{cc_script} 占位由本函数 .format() 回填,值逐字等价)。"""
+    Task 4 词表数据化: 指引话术数据源 = 墙 reminders.cc_guide
+    ({intl}/{cc_script} 占位由本函数 .format() 回填,值逐字等价)。
+    C-1(终审): wall=None = 模块级默认墙;manager 一律传租户墙。"""
     if not wa_state or not cc_completion.cc_due(wa_state):
         return ""
+    w = wall if wall is not None else DEFAULT_WALL
     if str(wa_state.get("e164") or ""):
         intl = _cc_intl(wa_state)
-        return DEFAULT_WALL.reminders["cc_guide"]["echo"].format(
-            intl=intl, cc_script=_cc_script(wa_state, lang))
-    return DEFAULT_WALL.reminders["cc_guide"]["ask"].format(
-        cc_script=_cc_script(wa_state, lang))
+        return w.reminders["cc_guide"]["echo"].format(
+            intl=intl, cc_script=_cc_script(wa_state, lang, wall=wall))
+    return w.reminders["cc_guide"]["ask"].format(
+        cc_script=_cc_script(wa_state, lang, wall=wall))
 
 
 def _contact_completion_note(
     raw_info: dict | None, confirmed: dict | None, suspect: dict | None,
     lang: str = "en", email_ask: bool = True,
+    wall: CompiledWall | None = None,
 ) -> tuple[str, list[str]]:
     """邮箱轻补提醒(spec §4.2/§4.3b,2026-10-08 判定树设计收编)。
     纯函数,返回(注入文案,计数键)。
@@ -208,6 +229,9 @@ def _contact_completion_note(
     suspect 仅为与调用处画像判定同签名,规则只看 confirmed(spec §4.2)。
     email_ask=False(Task 9: contact.email_light_ask 开关)→ 恒空,不注入
     也不产计数键。
+    C-1(终审): wall=None = 模块级默认墙(_EMAIL_NOTE_* 常量即该默认视图);
+    manager 一律传租户墙(流式调用处与本函数在 reminder 内的调用同参,
+    写回与注入永不漂移)。
     """
     if not email_ask:
         return "", []
@@ -222,7 +246,9 @@ def _contact_completion_note(
         return "", []
     if not cc_completion.cc_settled((info.get("_cc") or {}).get("whatsapp")):
         return "", []   # whatsapp 确认流未落定: 邮箱轻补串行等待
-    return (_EMAIL_NOTE_ZH if lang == "zh" else _EMAIL_NOTE_EN), ["_email_asked"]
+    w = wall if wall is not None else DEFAULT_WALL
+    return ((w.reminders["email_light"]["zh"] if lang == "zh"
+             else w.reminders["email_light"]["en"]), ["_email_asked"])
 
 
 def _cc_pre_turn(raw_info: dict, confirmed: dict, ip_iso: str | None) -> bool:
@@ -369,6 +395,11 @@ class ConversationManager:
         # 的 PromptProvider + compiled_wall 组装($DEFLECTION_POOL 单一事实源
         # 注入)。Task 9: AI 凭据/config 全部经 runtime.config.data 接线。
         self.runtime = runtime
+        # C-1(终审审查): 执行侧墙接线——过滤/替换/提醒全走租户编译墙,
+        # 与 AIService 的 $DEFLECTION_POOL 同源(runtime.compiled_wall);
+        # 模块级 DEFAULT_WALL 仅供无租户上下文的直调默认(boundary_filter
+        # 出厂视图 + 本模块旧测试锚定),会话流内一律本墙。
+        self._wall = runtime.compiled_wall
         self.ai = ai_service or AIService(
             prompts=runtime.prompts,
             wall=runtime.compiled_wall,
@@ -499,8 +530,8 @@ class ConversationManager:
 
         context_reminder = self._build_context_reminder(
             effective_raw, content, history, source=conv.source.value,
-            contact_flags=self._contact_flags)
-        if CONTACT_ASK_TRIGGER_RE.search(content or ""):
+            contact_flags=self._contact_flags, wall=self._wall)
+        if self._wall.triggers["contact_ask"].search(content or ""):
             logger.info("Contact-ask reminder fired (source=%s)", conv.source.value)
 
         # Task 3(spec §4.3/§6-1): 一次性计数写回 + _cc 状态同事务落库——
@@ -512,7 +543,7 @@ class ConversationManager:
         _, completion_marks = _contact_completion_note(
             effective_raw, _confirmed_now, _suspect_now,
             "zh" if _CJK_RE.search(content or "") else "en",
-            email_ask=self._contact_flags["email_light_ask"])
+            email_ask=self._contact_flags["email_light_ask"], wall=self._wall)
         if completion_marks or _cc_changed:
             raw = dict(profile.raw_info or {})
             for key in completion_marks:
@@ -577,12 +608,13 @@ class ConversationManager:
             pending += token
             complete, pending = split_sentences(pending)
             for sentence in complete:
-                if violates_boundary(sentence):
+                if violates_boundary(sentence, wall=self._wall):
                     violated = True
                     dirty = sentence
                     violation = "boundary"
                     break
-                if no_contact_value(sentence, allow_number=cc_allow_number):
+                if no_contact_value(sentence, allow_number=cc_allow_number,
+                                    wall=self._wall):
                     violated = True
                     dirty = sentence
                     violation = "no_contact"
@@ -590,7 +622,8 @@ class ConversationManager:
                 # Task 5b 再索要墙: 已留资时索要句静默丢弃(丢句不掐断,
                 # 其余句子照常流出——软于两墙的掐断+替换);
                 # 渠道精确化: 只丢索要已在手渠道的句子
-                if has_contact and asks_contact(sentence, _confirmed_channels):
+                if has_contact and asks_contact(sentence, _confirmed_channels,
+                                                wall=self._wall):
                     logger.warning("Re-ask wall dropped sentence (conv=%s): %r",
                                    conv.id, sentence[:60])
                     dropped_ask = True
@@ -604,15 +637,17 @@ class ConversationManager:
             if violated:
                 break  # 提前掐断生成,省 token
         if not violated and pending:
-            if violates_boundary(pending):
+            if violates_boundary(pending, wall=self._wall):
                 violated = True
                 dirty = pending
                 violation = "boundary"
-            elif no_contact_value(pending, allow_number=cc_allow_number):
+            elif no_contact_value(pending, allow_number=cc_allow_number,
+                                  wall=self._wall):
                 violated = True
                 dirty = pending
                 violation = "no_contact"
-            elif has_contact and asks_contact(pending, _confirmed_channels):
+            elif has_contact and asks_contact(pending, _confirmed_channels,
+                                              wall=self._wall):
                 # Task 5b 再索要墙: 尾巴是索要句 → 静默丢弃(不流出不掐断)
                 logger.warning("Re-ask wall dropped tail (conv=%s): %r",
                                conv.id, pending[:60])
@@ -636,21 +671,26 @@ class ConversationManager:
                     # NO_CONTACT_ALREADY——GLM 有 3/3 回显倾向,该路径概率高,
                     # 客户该收到确认/追问而非"稍后联系"虚假承诺;
                     # 价格墙路径(violation=="boundary")不变
-                    closing = _cc_wall_template(_wa_cc, _cc_lang_of(content))
+                    closing = _cc_wall_template(_wa_cc, _cc_lang_of(content),
+                                                wall=self._wall)
                 else:
-                    # Task 9 收口: 视图直取 DEFAULT_WALL.no_contact(单一
-                    # 事实源),不再经 boundary_filter 的模块级常量转引
-                    closing = (DEFAULT_WALL.no_contact["already"] if has_contact
-                               else (DEFAULT_WALL.no_contact["app"]
+                    # Task 9 收口: 视图直取墙 no_contact(单一事实源),
+                    # 不经 boundary_filter 的模块级常量转引;C-1(终审):
+                    # 取租户墙——替换文案各自来自自己的池
+                    _w = self._wall
+                    closing = (_w.no_contact["already"] if has_contact
+                               else (_w.no_contact["app"]
                                      if conv.source.value == "app"
-                                     else DEFAULT_WALL.no_contact["web"]))
+                                     else _w.no_contact["web"]))
             else:
                 logger.warning(
                     "Boundary filter intercepted sentence (conv=%s): %r | tail=%r",
                     conv.id, dirty[:80], (pending or "")[:40])
-                closing = (pivot_for(content, has_contact) if emitted
+                closing = (pivot_for(content, has_contact, wall=self._wall)
+                           if emitted
                            else deflection_for(content, _boundary_rotate(),
-                                               has_contact=has_contact))
+                                               has_contact=has_contact,
+                                               wall=self._wall))
             # Minor-5(审查修复): 已流出前缀与替换句之间补一个空格分隔
             # ("...xxxx? Got it —"——英文衔接无空格会粘连);CC/触发轮缓冲
             # 未流出,emitted 恒空,天然不受影响。空格随 closing 一起流出,
@@ -690,7 +730,8 @@ class ConversationManager:
                 logger.warning(
                     "CC enforcement wall replaced reply (conv=%s): %r",
                     conv.id, cc_full[:60])
-                cc_template = _cc_wall_template(_wa_cc, cc_lang)
+                cc_template = _cc_wall_template(_wa_cc, cc_lang,
+                                                wall=self._wall)
                 for i in range(0, len(cc_template), 40):
                     yield {"type": "stream_chunk",
                            "content": cc_template[i:i + 40]}
@@ -702,7 +743,7 @@ class ConversationManager:
         if not violated and dropped_ask and not emitted.strip():
             logger.warning("Re-ask wall dropped entire reply (conv=%s), "
                            "falling back to closing", conv.id)
-            fallback = DEFAULT_WALL.no_contact["already"]
+            fallback = self._wall.no_contact["already"]
             for i in range(0, len(fallback), 40):
                 yield {"type": "stream_chunk",
                        "content": fallback[i:i + 40]}
@@ -1141,24 +1182,29 @@ class ConversationManager:
     def _build_context_reminder(raw_info: dict, latest_msg: str,
                                 history: list[dict] | None = None,
                                 source: str = "",
-                                contact_flags: dict | None = None) -> str:
+                                contact_flags: dict | None = None,
+                                wall: CompiledWall | None = None) -> str:
         """Build a context reminder from extracted profile info.
         Task 4 词表数据化: 六类轮中提醒话术(boundary/contact_ask/
         language_follow/lead_given/cc_guide/email_light)的数据源 =
-        DEFAULT_WALL.reminders(seed_profiles/jiancai/wall.json);
+        墙 reminders(seed_profiles/<slug>/wall.json);
         {shown}/{no_contact_web}/{no_contact_already} 占位由本函数
         .format() 回填,值与数据化前逐字等价。
         contact_flags(Task 9): contact 行为开关——cc_completion=False 跳过
         cc 确认指引,email_light_ask=False 跳过邮箱轻补;None = 全开
-        (直调本函数的旧测试语义)。"""
+        (直调本函数的旧测试语义)。
+        wall(C-1 终审): 租户编译墙(触发词/提醒话术/no_contact 全走它);
+        None = 模块级默认墙(直调/旧测试语义——BOUNDARY_TRIGGER_RE 等
+        模块常量即该默认视图)。"""
+        w = wall if wall is not None else DEFAULT_WALL
         # 职能边界提醒(评审#7: 前置到早退之前,首条消息画像为空也要生效)
         boundary_note = ""
-        if BOUNDARY_TRIGGER_RE.search(latest_msg or ""):
-            if BUDGET_TRIGGER_RE.search(latest_msg or ""):
+        if w.triggers["boundary"].search(latest_msg or ""):
+            if w.triggers["budget"].search(latest_msg or ""):
                 # 评审#9: 预算求推荐 → 预算路径(推荐+反问+引导),不一刀切要号
-                boundary_note = DEFAULT_WALL.reminders["boundary"]["budget"]
+                boundary_note = w.reminders["boundary"]["budget"]
             else:
-                boundary_note = DEFAULT_WALL.reminders["boundary"]["normal"]
+                boundary_note = w.reminders["boundary"]["normal"]
 
         # 联系方式状态前移(供 boundary/contact 提醒共用判定)
         contacts_all = normalize_contacts(raw_info.get("联系方式"))
@@ -1169,23 +1215,23 @@ class ConversationManager:
         # 冒烟实证: 抽象指令("用App话术")GLM 会无视并背 prompt 里更显眼的 web 脚本——
         # 提醒必须携带该分支的逐字话术,不让模型自己挑。
         contact_note = ""
-        if CONTACT_ASK_TRIGGER_RE.search(latest_msg or ""):
+        if w.triggers["contact_ask"].search(latest_msg or ""):
             if source == "app":
                 if confirmed or suspect:
                     # 已留资: 索要我方联系方式 → 逐字"经理主动联系"版,
                     # 与"禁止再索要"提醒不再打架
                     # Task 3 收尾: 第三处禁令同步改写(禁令集完整)——与前两处
                     # 同语义,只禁已提供过的渠道,缺失渠道放行轻补一次
-                    contact_note = DEFAULT_WALL.reminders["contact_ask"]["app_already"]
+                    contact_note = w.reminders["contact_ask"]["app_already"]
                 else:
-                    contact_note = DEFAULT_WALL.reminders["contact_ask"]["app"]
+                    contact_note = w.reminders["contact_ask"]["app"]
             else:
-                contact_note = DEFAULT_WALL.reminders["contact_ask"]["web"].format(
-                    no_contact_web=DEFAULT_WALL.no_contact["web"])
+                contact_note = w.reminders["contact_ask"]["web"].format(
+                    no_contact_web=w.no_contact["web"])
 
         # 语言跟随提醒(Task 11): 与 boundary/contact 提醒同挂载位——
         # 首条消息画像为空也要生效(中文客户开场即漂英文是实测重灾区)
-        lang_note = _lang_note_for(latest_msg)
+        lang_note = _lang_note_for(latest_msg, wall=wall)
 
         if not raw_info:
             # 空画像:仅返回边界/联系/语言提醒(可能为空串)
@@ -1213,7 +1259,7 @@ class ConversationManager:
         completion_note, _marks = _contact_completion_note(
             raw_info, confirmed, suspect,
             lang="zh" if _CJK_RE.search(latest_msg or "") else "en",
-            email_ask=flags["email_light_ask"])
+            email_ask=flags["email_light_ask"], wall=wall)
 
         # Task 4(spec §4 回显确认轮): cc 等待态注入确认句指引——有 e164 用
         # 回显模板(逐字 intl),no_ip 用 ASK 追问句;与邮箱轻补 cc_settled
@@ -1221,30 +1267,31 @@ class ConversationManager:
         # attempts 管理(轮末 transit 递增),提醒每轮等待态都在场。
         # P3: 语言路由扩 ES(_cc_lang_of),西语轮注入西语模板。
         cc_note = (_cc_confirm_note(
-            (raw_info.get("_cc") or {}).get("whatsapp"), _cc_lang_of(latest_msg))
+            (raw_info.get("_cc") or {}).get("whatsapp"), _cc_lang_of(latest_msg),
+            wall=wall)
             if flags["cc_completion"] else "")
 
         # 联系方式专项(D6) 与 职能边界(D6联动)
         if confirmed:
             shown = ", ".join(
                 f"{k}: {' '.join(str(v).split())[:60]}" for k, v in confirmed.items())
-            # 逐字话术直接引用 DEFAULT_WALL.no_contact["already"](Task 9 收口:
+            # 逐字话术直接引用墙 no_contact["already"](Task 9 收口:
             # 视图直取单一事实源)——与零号码墙替换句同源,三处"经理主动联系"
             # 话术不漂移
             # Task 3 防自锁改写(spec 4.1-6): 原"绝对禁止再索要任何联系方式！"
             # 会抑制缺失渠道轻补——只禁已提供过的渠道,缺失渠道放行一次
-            parts.append(DEFAULT_WALL.reminders["lead_given"]["forbidden"].format(
+            parts.append(w.reminders["lead_given"]["forbidden"].format(
                 shown=shown))
-            parts.append(DEFAULT_WALL.reminders["lead_given"]["respond_script"].format(
-                no_contact_already=DEFAULT_WALL.no_contact["already"]))
+            parts.append(w.reminders["lead_given"]["respond_script"].format(
+                no_contact_already=w.no_contact["already"]))
             if boundary_note:
                 parts.append(boundary_note)
                 # Task 3 防自锁改写: 尾部补缺失渠道豁免(与上方改写同因)
-                parts.append(DEFAULT_WALL.reminders["lead_given"]["boundary_tail"])
+                parts.append(w.reminders["lead_given"]["boundary_tail"])
         elif suspect:
             shown = ", ".join(
                 f"{k}: {' '.join(str(v).split())[:60]}" for k, v in suspect.items())
-            parts.append(DEFAULT_WALL.reminders["lead_given"]["suspect"].format(
+            parts.append(w.reminders["lead_given"]["suspect"].format(
                 shown=shown))
             if boundary_note:
                 parts.append(boundary_note)

requirements.txt

@@ -2,7 +2,6 @@ fastapi>=0.115.0
 uvicorn[standard]>=0.34.0
 sqlalchemy[asyncio]>=2.0.0
 aiosqlite>=0.20.0
-alembic>=1.14.0
 zhipuai>=2.1.5
 aiosmtplib>=3.0.0
 httpx>=0.27.0

seed_profiles/auto/tenant.json

@@ -17,6 +17,10 @@
       "welcome": "Hey! What kind of car are you looking for?",
       "topics": ["Get Quotation", "Distributor Cooperation"]
     },
+    "email": {
+      "sender_name": "XYH",
+      "notify_emails": ["18339827479@163.com"]
+    },
     "push": {
       "prefix": "ai-kefu"
     },

seed_profiles/jiancai/tenant.json

@@ -26,6 +26,10 @@
       "welcome": "What kind of piping project are you working on?",
       "topics": ["Get Quotation", "Pipe Specifications", "Delivery & Shipping"]
     },
+    "email": {
+      "sender_name": "HESUPIPE",
+      "notify_emails": ["hesu@hesuu.com"]
+    },
     "push": {
       "prefix": "ai-kefu-jc"
     },

tests/conftest.py

@@ -107,6 +107,22 @@ def build_jiancai_runtime():
     return _JIANCAI_RUNTIME
 
 
+_AUTO_RUNTIME = None
+
+
+def build_auto_runtime():
+    """(非夹具)auto TenantRuntime——C-1 终审回归锁用(两租户墙差异断言),
+    纪律同 build_jiancai_runtime: 只走加载期,无 engine 无句柄,
+    模块级缓存单例跨测试复用。"""
+    global _AUTO_RUNTIME
+    if _AUTO_RUNTIME is None:
+        reg = TenantRegistry.load_from_dir(
+            SEED_DIR, data_dir=REPO_ROOT / "tenants")
+        _AUTO_RUNTIME = next(
+            rt for rt in reg.runtimes() if rt.config.slug == "auto")
+    return _AUTO_RUNTIME
+
+
 # Task 7: app_client 默认域名——既有测试不带 host 头,默认落 jiancai
 # (与 5b 之前 default_runtime 兜底的语义一致,既有断言零改动)
 JIANCAI_HOST = "chat2.agent.hnxinyouhe.com"

tests/test_tenant_config_wiring.py

@@ -103,6 +103,24 @@ def test_send_lead_email_uses_tenant_sender_and_recipients(monkeypatch):
     assert str(captured["msg"]["To"]) == "x@t.com, y@t.com"
 
 
+def test_seed_tenants_ship_email_group(monkeypatch):
+    """I-1(终审): 两租户 seed 自带 email 组——resolve_email_settings 各自
+    返回正确收件人/发件名(auto: XYH→163 信箱; jiancai: HESUPIPE→hesuu.com),
+    不再依赖 env NOTIFY_EMAIL 平台兜底(patch 成异值证明 seed 槽胜出)。
+    收件人取值依据 brand_values.json: auto=notify_email 环境实际值,
+    jiancai=品牌邮箱(notify_email 记录的是 notify@example.com 占位)。
+    From 头渲染时外层再包【】(send_* 的 msg["From"] 拼法),seed 存裸名。"""
+    monkeypatch.setattr(settings, "NOTIFY_EMAIL", "env-notify@x.com")
+    from tests.conftest import build_auto_runtime
+    au = email_service.resolve_email_settings(build_auto_runtime().config.data)
+    jc_r = email_service.resolve_email_settings(
+        build_jiancai_runtime().config.data)
+    assert au["sender_name"] == "XYH"
+    assert au["notify_emails"] == ["18339827479@163.com"]
+    assert jc_r["sender_name"] == "HESUPIPE"
+    assert jc_r["notify_emails"] == ["hesu@hesuu.com"]
+
+
 # ---- push ----------------------------------------------------------------
 
 def _patch_push_env(monkeypatch):

tests/test_wall_execution_tenant.py

@@ -0,0 +1,121 @@
+"""C-1(终审审查)回归锁: 墙的执行侧必须走租户管道。
+
+修复前缺陷: conversation.py 执行侧(过滤/替换/提醒)全走模块级
+DEFAULT_WALL(jiancai 出厂),runtime.compiled_wall 全库只有 AIService 的
+$DEFLECTION_POOL 一处消费——auto 租户会话实际跑的是 jc 墙(多拦/替换
+文案串租户)。本文件用两租户真 runtime + 脚本化假 AI 端到端断言:
+
+- 价格墙: jc 独有 pattern([€£]\\s?\\d,auto 词表无此组)命中的句子在
+  jiancai 被拦+替换(收尾来自 jc pivots 池),在 auto 放行逐字流出;
+- 零号码墙: 同一邮箱违规句两租户都拦,替换句精确等于各自墙的
+  no_contact["web"](jc "within 12 hours" vs auto "replies fast!");
+- 提醒侧: contact_ask 触发消息的注入 context 携带各自墙的
+  no_contact["web"](web 模板占位 {no_contact_web} 回填)。
+"""
+import asyncio
+
+from sqlalchemy import select
+
+from app.engine.models import Message, MessageRole
+from app.engine.services.conversation import ConversationManager
+from tests.conftest import build_auto_runtime, build_jiancai_runtime
+
+# 判别句选取依据(两墙 diff 实证,勿随手改文案——判别力依赖词表差异):
+# - PRICE_REPLY 撞 jc 独有的 [€£]\s?\d 组(auto 词表无),零号码两墙都不拦;
+# - EMAIL_REPLY 撞零号码墙 any_email(两墙同拦,替换文案不同);
+# - 末句无尾随空白 → 以 pending 尾巴形态裁决,走同一条最终检查路径。
+PRICE_REPLY = "Thanks for asking. The total is €500 for one set."
+EMAIL_REPLY = "Sure, you can also email me at sales@example.com anytime."
+USER_MSG = "do you ship to Nigeria?"
+
+
+class _ScriptedAI:
+    """固定回复脚本桩: 小块 token 流出 reply(走真实 split_sentences/
+    逐句门控路径),捕获注入的 context(提醒侧断言用)。"""
+
+    def __init__(self, reply):
+        self._reply = reply
+        self.captured_context = None
+
+    async def stream_chat(self, messages, context=""):
+        self.captured_context = context
+        for i in range(0, len(self._reply), 7):
+            yield self._reply[i:i + 7]
+
+    async def extract_info(self, messages, existing_info):
+        return {}
+
+    async def evaluate(self, profile, recent_messages):
+        return {"intent_level": "pending", "reasoning": "",
+                "contact_info": {}, "info_sufficient": False}
+
+
+def _send(async_db, runtime, reply, user_msg=USER_MSG):
+    """一轮会话 → (落库的助手回复, 注入 context)。落库文本与流出统一
+    (stream_process_message 契约),取 Message 表即可代表客户所见。"""
+
+    async def run():
+        async with async_db() as db:
+            ai = _ScriptedAI(reply)
+            mgr = ConversationManager(db, runtime=runtime, ai_service=ai)
+            conv = await mgr.create_conversation(source=None)  # standalone
+            async for _ in mgr.stream_process_message(conv.session_id,
+                                                      user_msg):
+                pass
+            msgs = (await db.execute(
+                select(Message).where(Message.conversation_id == conv.id)
+                .order_by(Message.created_at))).scalars().all()
+            replies = [m.content for m in msgs
+                       if m.role == MessageRole.assistant]
+            return (replies[-1] if replies else ""), ai.captured_context
+    return asyncio.run(run())
+
+
+def test_price_wall_pattern_is_tenant_scoped(async_db):
+    """jc 独有 [€£] 价格 pattern: jiancai 拦截替换(€ 数字不流出,收尾 =
+    jc pivots 池首句);auto 放行逐字流出——同一引擎代码,墙随租户。"""
+    jc = build_jiancai_runtime()
+    au = build_auto_runtime()
+    # 判别前提(词表趋同护栏): 该句必须只撞 jc 墙,否则本测试失去判别力
+    from app.engine.services.boundary_filter import violates_boundary
+    assert violates_boundary(PRICE_REPLY, wall=jc.compiled_wall) is True
+    assert violates_boundary(PRICE_REPLY, wall=au.compiled_wall) is False
+
+    jc_reply, _ = _send(async_db, jc, PRICE_REPLY)
+    assert "€500" not in jc_reply
+    # 已流出干净前缀("Thanks for asking.")→ 违规后接 pivot 过渡句收尾
+    assert jc_reply.endswith(jc.compiled_wall.pivots["en"][0])
+
+    au_reply, _ = _send(async_db, au, PRICE_REPLY)
+    assert au_reply == PRICE_REPLY  # auto 墙无此 pattern → 整句放行
+
+
+def test_no_contact_replacement_copy_from_tenant_pool(async_db):
+    """零号码墙(两租户同拦同一邮箱句): 替换句精确等于各自墙的
+    no_contact["web"](standalone 源+未留资 → web 变体)。修复前 auto
+    会话收的是 jc 文案(12 hours),本测试即失败。"""
+    jc = build_jiancai_runtime()
+    au = build_auto_runtime()
+    assert (jc.compiled_wall.no_contact["web"]
+            != au.compiled_wall.no_contact["web"])  # 判别前提
+
+    jc_reply, _ = _send(async_db, jc, EMAIL_REPLY)
+    au_reply, _ = _send(async_db, au, EMAIL_REPLY)
+    assert jc_reply == jc.compiled_wall.no_contact["web"]
+    assert au_reply == au.compiled_wall.no_contact["web"]
+
+
+def test_contact_ask_reminder_uses_tenant_wall_text(async_db):
+    """提醒侧同锁: contact_ask 触发消息的注入 context 携带各自墙的
+    no_contact["web"] 文案(web 模板 {no_contact_web} 回填),互不渗漏。"""
+    jc = build_jiancai_runtime()
+    au = build_auto_runtime()
+    jc_web = jc.compiled_wall.no_contact["web"]
+    au_web = au.compiled_wall.no_contact["web"]
+
+    _, jc_ctx = _send(async_db, jc, "ok.", user_msg="how can I contact you?")
+    _, au_ctx = _send(async_db, au, "ok.", user_msg="how can I contact you?")
+    assert jc_web in jc_ctx
+    assert au_web in au_ctx
+    assert au_web not in jc_ctx
+    assert jc_web not in au_ctx