feat: 1、Vo包装识别增强 2、去除 位置&类型 的色块

.gitea/config/serialization-schema-check-config.yaml

@@ -1,10 +1,10 @@
 # ============================================================
-# 缓存序列化结构变更检测 — 业务仓库配置
+# 序列化结构变更检测 — 业务仓库配置
 # ============================================================
 # 说明：
 #   - 本配置文件为业务覆盖配置，会与 jar 内 default-config.yaml 深度合并
 #   - 未声明的项沿用工具内置默认值（忽略规则、检测模式等）
-#   - 当前实现以 Redis 写入检测为主，后续可扩展其他缓存
+#   - 当前以 Redis 缓存写入检测为主；MQ（RocketMQ/Kafka）方案见 docs/MQ序列化结构检测方案.md
 
 # 总开关    true-执行检测   false-跳过检测（流水线直接通过，不发通知）
 enabled: true
@@ -17,7 +17,7 @@ notify:
   enabled: true
   webhook_url: https://qyapi.weixin.qq.com/cgi-bin/webhook/send?key=fa14f0b3-e01a-40f6-96bd-e18beb94e85e
   notify_on_clean: false
-  title_prefix: "【缓存结构变更】"
+  title_prefix: "【序列化结构变更】"
 
 # 观察期：先只扫描 jnpf-tenant 模块，稳定后改为 include_modules: []
 include_modules:

.gitea/workflows/serialization-schema-check.yaml

@@ -1,17 +1,17 @@
-name: 缓存序列化结构检查
+name: 序列化结构检查
-run-name: ${{ gitea.actor }}的缓存结构检查
+run-name: ${{ gitea.actor }}的序列化结构检查
 
 on:
   push:
   workflow_dispatch:
 
 env:
-  # cache-schema-checker 私库坐标：com.codechecker:cache-schema-checker:1.0.0
+  # serialization-schema-checker 私库坐标：com.codechecker:serialization-schema-checker:1.0.0
-  CACHE_SCHEMA_CHECKER_VERSION: "1.0.0"
+  SERIALIZATION_SCHEMA_CHECKER_VERSION: "1.0.0"
-  CACHE_SCHEMA_CHECKER_REPO_URL: "http://192.168.3.25:18081/nexus/repository/maven-releases"
+  SERIALIZATION_SCHEMA_CHECKER_REPO_URL: "http://192.168.3.25:18081/nexus/repository/maven-releases"
 
 jobs:
-  cache-schema-check:
+  serialization-schema-check:
     if: ${{ gitea.ref != 'refs/heads/pre' && gitea.ref != 'refs/heads/dev' && gitea.ref != 'refs/heads/master-2.0' }}
     runs-on: jdk11
     steps:
@@ -27,30 +27,30 @@ jobs:
           git clone --depth 1 --single-branch --branch "${BRANCH}" "${REPO_URL}" .
           git checkout -B "${BRANCH}" "${NEW_SHA}"
 
-          echo "${NEW_SHA}" > /tmp/cache-schema-new-sha.txt
+          echo "${NEW_SHA}" > /tmp/serialization-schema-new-sha.txt
 
       - name: 检查配置文件
         run: |
-          if [ ! -f .gitea/config/cache-schema-check-config.yaml ]; then
+          if [ ! -f .gitea/config/serialization-schema-check-config.yaml ]; then
-            echo "错误: 缺少 .gitea/config/cache-schema-check-config.yaml"
+            echo "错误: 缺少 .gitea/config/serialization-schema-check-config.yaml"
             exit 1
           fi
           # 顶层总开关 enabled: false 时跳过后续步骤（与 notify.enabled 区分，仅匹配行首）
-          if grep -Eq '^enabled:[[:space:]]*false([[:space:]]|#|$)' .gitea/config/cache-schema-check-config.yaml; then
+          if grep -Eq '^enabled:[[:space:]]*false([[:space:]]|#|$)' .gitea/config/serialization-schema-check-config.yaml; then
-            echo "总开关 enabled=false，跳过缓存结构检查"
+            echo "总开关 enabled=false，跳过序列化结构检查"
-            touch /tmp/cache-schema-check.skip
+            touch /tmp/serialization-schema-check.skip
           fi
 
-      - name: 从 Nexus 私库下载 cache-schema-checker
+      - name: 从 Nexus 私库下载 serialization-schema-checker
         run: |
-          if [ -f /tmp/cache-schema-check.skip ]; then
+          if [ -f /tmp/serialization-schema-check.skip ]; then
             echo "总开关已关闭，跳过下载"
             exit 0
           fi
 
-          GROUP_PATH="com/codechecker/cache-schema-checker"
+          GROUP_PATH="com/codechecker/serialization-schema-checker"
-          JAR_NAME="cache-schema-checker-${CACHE_SCHEMA_CHECKER_VERSION}.jar"
+          JAR_NAME="serialization-schema-checker-${SERIALIZATION_SCHEMA_CHECKER_VERSION}.jar"
-          JAR_URL="${CACHE_SCHEMA_CHECKER_REPO_URL}/${GROUP_PATH}/${CACHE_SCHEMA_CHECKER_VERSION}/${JAR_NAME}"
+          JAR_URL="${SERIALIZATION_SCHEMA_CHECKER_REPO_URL}/${GROUP_PATH}/${SERIALIZATION_SCHEMA_CHECKER_VERSION}/${JAR_NAME}"
           JAR_PATH="/tmp/${JAR_NAME}"
 
           echo "下载: ${JAR_URL}"
@@ -72,25 +72,25 @@ jobs:
 
       - name: 验证 JDK
         run: |
-          if [ -f /tmp/cache-schema-check.skip ]; then
+          if [ -f /tmp/serialization-schema-check.skip ]; then
             echo "总开关已关闭，跳过"
             exit 0
           fi
           echo "Java: $(java -version 2>&1 | head -1)"
 
-      - name: 执行缓存序列化结构检测
+      - name: 执行序列化结构检测
         env:
           # push 前 tip；新分支首次 push 时为全 0。workflow_dispatch 可能为空，下方会回退。
           PUSH_BEFORE: ${{ gitea.event.before }}
           # push 事件携带的 commits 列表（JSON）；用于准确统计 commit 数（不受浅克隆影响）
           PUSH_COMMITS_JSON: ${{ toJson(gitea.event.commits) }}
         run: |
-          if [ -f /tmp/cache-schema-check.skip ]; then
+          if [ -f /tmp/serialization-schema-check.skip ]; then
             echo "总开关已关闭，跳过检测"
             exit 0
           fi
 
-          NEW_SHA=$(cat /tmp/cache-schema-new-sha.txt)
+          NEW_SHA=$(cat /tmp/serialization-schema-new-sha.txt)
           OLD_SHA="${PUSH_BEFORE}"
 
           # 新分支首次 push（before 全 0）→ 跳过
@@ -199,8 +199,8 @@ jobs:
 
           COMMIT_TIME=$(git log -1 --format=%cd --date=format:'%Y-%m-%d %H:%M:%S' "${NEW_SHA}")
 
-          java -jar "/tmp/cache-schema-checker-${CACHE_SCHEMA_CHECKER_VERSION}.jar" \
+          java -jar "/tmp/serialization-schema-checker-${SERIALIZATION_SCHEMA_CHECKER_VERSION}.jar" \
-            --config .gitea/config/cache-schema-check-config.yaml \
+            --config .gitea/config/serialization-schema-check-config.yaml \
             --repo-root . \
             --old-sha "$OLD_SHA" \
             --new-sha "$NEW_SHA" \

.idea/compiler.xml

@@ -1,4 +1,4 @@
-<?xml version="1.0" encoding="UTF-8"?>
+<?xml version="1.0" encoding="UTF-8"?>
 <project version="4">
   <component name="CompilerConfiguration">
     <annotationProcessing>
@@ -6,11 +6,11 @@
         <sourceOutputDir name="target/generated-sources/annotations" />
         <sourceTestOutputDir name="target/generated-test-sources/test-annotations" />
         <outputRelativeToContentRoot value="true" />
-        <module name="cache-schema-checker" />
+        <module name="serialization-schema-checker" />
       </profile>
     </annotationProcessing>
     <bytecodeTargetLevel>
-      <module name="cache-schema-checker" target="11" />
+      <module name="serialization-schema-checker" target="11" />
     </bytecodeTargetLevel>
   </component>
 </project>

docs/CI集成说明.md

@@ -1,4 +1,4 @@
-# 缓存序列化结构检测 — CI 集成说明
+# 序列化结构检测 — CI 集成说明
 
 ---
 
@@ -11,7 +11,7 @@ Gitea Actions 触发
     ↓
 浅克隆业务仓库 tip（depth=1）+ 按需取 push 前 tip（before）
     ↓
-从 Nexus 下载 cache-schema-checker.jar
+从 Nexus 下载 serialization-schema-checker.jar
     ↓
 java -jar 执行（对比 before → after，累计 diff）
     ↓
@@ -46,7 +46,7 @@ mode=block 且含任意结构变更 → exit 1（流水线失败）
 |----|------|
 | Gitea Runner | 标签 `jdk11`，已安装 Java 11 |
 | Nexus 私库 | 可访问 `http://192.168.3.25:18081/nexus/repository/maven-releases` |
-| 工具 JAR | `com.codechecker:cache-schema-checker:1.0.0` 已发布 |
+| 工具 JAR | `com.codechecker:serialization-schema-checker:1.0.0` 已发布 |
 
 ---
 
@@ -58,18 +58,18 @@ mode=block 且含任意结构变更 → exit 1（流水线失败）
 jnpf-java-cloud/
 ├── .gitea/
 │   ├── workflows/
-│   │   └── cache-schema-check.yaml      # 流水线
+│   │   └── serialization-schema-check.yaml      # 流水线
 │   └── config/
-│       └── cache-schema-check-config.yaml  # 检测配置
+│       └── serialization-schema-check-config.yaml  # 检测配置
 ```
 
-请以本仓库 `.gitea/workflows/cache-schema-check.yaml` 为模板同步到业务仓。
+请以本仓库 `.gitea/workflows/serialization-schema-check.yaml` 为模板同步到业务仓。
 
 ---
 
 ## 4. 流水线模板（要点）
 
-完整可运行版本见：`.gitea/workflows/cache-schema-check.yaml`。
+完整可运行版本见：`.gitea/workflows/serialization-schema-check.yaml`。
 
 核心逻辑摘要：
 
@@ -84,7 +84,7 @@ git fetch --depth 1 origin "$OLD_SHA"   # 优先
 # 或 git fetch --deepen N               # 兜底
 
 # 4) 执行
-java -jar cache-schema-checker-1.0.0.jar \
+java -jar serialization-schema-checker-1.0.0.jar \
   --old-sha "$OLD_SHA" \
   --new-sha "$NEW_SHA" \
   ...
@@ -98,7 +98,7 @@ java -jar cache-schema-checker-1.0.0.jar \
 |--------|------|------|
 | `demo.yaml` (AI代码质量分析) | AI Code Review | 并行，互不影响 |
 | `code-check` (CodeChecker) | 通用变更检测 | **同模式**，可并列执行 |
-| `cache-schema-check` | 缓存结构检测 | 新增 |
+| `serialization-schema-check` | 缓存结构检测 | 新增 |
 
 
 ---
@@ -116,9 +116,9 @@ mvn clean deploy -DskipTests
 发布产物：
 
 ```text
-com/codechecker/cache-schema-checker/1.0.0/
+com/codechecker/serialization-schema-checker/1.0.0/
-  ├── cache-schema-checker-1.0.0.jar          # 可执行 fat-jar
+  ├── serialization-schema-checker-1.0.0.jar          # 可执行 fat-jar
-  └── cache-schema-checker-1.0.0.pom
+  └── serialization-schema-checker-1.0.0.pom
 ```
 
 ---
@@ -157,8 +157,8 @@ com/codechecker/cache-schema-checker/1.0.0/
 OLD_SHA=$(git rev-parse origin/$(git branch --show-current)~3)  # 示例：假设 ahead 3
 NEW_SHA=$(git rev-parse HEAD)
 
-java -jar /path/to/cache-schema-checker-1.0.0.jar \
+java -jar /path/to/serialization-schema-checker-1.0.0.jar \
-  --config .gitea/config/cache-schema-check-config.yaml \
+  --config .gitea/config/serialization-schema-check-config.yaml \
   --repo-root . \
   --old-sha "$OLD_SHA" \
   --new-sha "$NEW_SHA" \
@@ -169,3 +169,13 @@ java -jar /path/to/cache-schema-checker-1.0.0.jar \
 ```
 
 单 commit 自测仍可用 `--old-sha HEAD~1 --new-sha HEAD`。
+
+---
+
+## 10. 相关文档
+
+| 文档 | 说明 |
+|------|------|
+| [实施方案.md](./实施方案.md) | 缓存检测总体方案 |
+| [配置说明.md](./配置说明.md) | YAML 配置项 |
+| [MQ序列化结构检测方案.md](./MQ序列化结构检测方案.md) | MQ 消息体 Schema（RocketMQ + Kafka，方案已落地） |

docs/MQ序列化结构检测方案.md

@@ -0,0 +1,339 @@
+# MQ 消息体序列化结构变更检测 — 方案
+
+> 版本：v0.2  
+> 日期：2026-07-15  
+> 状态：**方案已落地（含 Kafka），开发未启动**  
+> 关联：复用 `serialization-schema-checker` 的 Schema 提取、Diff、企微通知与 CI 框架  
+> 业务样本仓：`jnpf-java-cloud`（**RocketMQ + Kafka**）
+
+---
+
+## 1. 背景与目标
+
+### 1.1 为什么要做
+
+业务同时使用 **RocketMQ** 与 **Kafka** 投递业务对象：
+
+| 中间件 | 典型写法 | 序列化要点 |
+|--------|----------|------------|
+| RocketMQ | `rocketMQTemplate.syncSend(topic:tag, dto)` | Spring MessageConverter（多为 Jackson）把对象变成消息体 |
+| Kafka | `kafkaTemplate.send(topic, vo|List)` | Spring `KafkaTemplate` + value serializer（多为 Json）编码对象 |
+| Kafka 消费 | `@KafkaListener` + `parseObject(message, Xxx.class)` | 常以 String 接收后再 Fastjson 反序列化 |
+
+当消息 DTO / VO **删字段、改类型、加包装层**时：
+
+- Topic / 重试队列里仍可能有**旧结构消息**
+- 新消费代码反序列化失败，或字段为空导致静默逻辑错误
+
+这与 Redis 缓存「残留旧 value」同一类问题：
+
+| | Redis | RocketMQ | Kafka |
+|--|-------|----------|-------|
+| 残留形态 | 未过期 key | Topic 积压 / 重试 | Topic 积压 / 消费 lag |
+| 路由标识 | key 模式 | **topic:tag** | **topic**（一般无 tag；动态后缀可归一 `*`） |
+| 典型序列化 | Fastjson 字符串或 Template 直写 | MessageConverter | Kafka JsonSerializer / 手写 JSON 字符串 |
+
+### 1.2 目标
+
+在 push 时静态分析 **消息体类型的序列化 Schema** 是否相对对比区间发生变更，覆盖 **RocketMQ + Kafka**，并复用现有企微通知 / notify|block 能力。
+
+### 1.3 非目标（本方案首版）
+
+- 不连接真实 Broker，不拉取积压消息做运行时校验
+- 不解析依赖 jar 内消息类型（仅本仓 `src/main/java`）
+- 不替代权限、幂等、消费失败重试等业务正确性检查
+- 不扫仅 Admin 建 Topic 的工具类（如 `KafkaTopicUtil`，无业务 body）
+- RabbitMQ 等若后续出现再扩展（当前仓以 RocketMQ / Kafka 为主）
+
+---
+
+## 2. 业务调研结论（jnpf-java-cloud）
+
+### 2.1 RocketMQ
+
+| 写法 | 出现情况 | 策略 |
+|------|----------|------|
+| `rocketMQTemplate.syncSend(dest, dto)` | 高（如钱包扣费） | **纳入** |
+| `asyncSend` / `syncSendOrderly` 等 | 中 | **纳入** |
+| `convertAndSend` | 视封装而定 | **纳入** |
+| 先 `JSON.toJSONString` 再发 String | 较低 | unwrap 后取类型 |
+| 只发 `String` / `byte[]` / `MessageExt` | 有 | **默认忽略** |
+
+样本（资金钱包）：
+
+```java
+rocketMQTemplate.syncSend(CapitalMqConstants.TOPIC + ":" + tag, req); // WalletDeductReq
+
+@RocketMQMessageListener(...)
+public class WalletDeductConsumer implements RocketMQListener<WalletDeductReq> { ... }
+```
+
+### 2.2 Kafka（已确认需纳入）
+
+| 写法 | 出现情况 | 策略 |
+|------|----------|------|
+| `kafkaTemplate.send(topic, dto)` | 中（值班食安项等） | **纳入** |
+| `kafkaTemplate.send(topic, List<Xxx>)` | 中（巡店食安项列表） | **纳入**（rootArray） |
+| `@KafkaListener` + `String` + `JSONObject.parseObject(..., Xxx.class)` | 有（数据分析中差评） | **读侧补强 MQ-R** |
+| `KafkaTopicUtil` 仅创建 Topic | 有（租户） | **忽略**（无消息体） |
+
+生产样本（巡店）：
+
+```java
+List<CheckItemDetailVo> thousandsData = ...;
+kafkaTemplate.send(topicBuilder.patrolStoreTopic(tenantId), thousandsData);
+```
+
+生产样本（值班）：
+
+```java
+KafkaTemplate<String, Object> kafkaTemplate;
+kafkaTemplate.send(topic, data); // CheckItemDetailVO
+```
+
+消费样本（数据分析）：
+
+```java
+@KafkaListener(topics = "ftb-evaluate-real-notification${...}", groupId = "...")
+public void handleMessage(String message) {
+    AddedMessageNotificationToVO vo = JSONObject.parseObject(message, AddedMessageNotificationToVO.class);
+}
+```
+
+动态 Topic（如按租户拼接）静态推断结果形如 `patrol-store-topic:*`，与 Redis key `*` 规则一致。
+
+### 2.3 「Key」等价物（destination）
+
+| 中间件 | 聚合键形态 | 来源 |
+|--------|------------|------|
+| RocketMQ | `topic:tag` | 字面量、常量、`TOPIC + ":" + TAG` |
+| Kafka | `topic` | 字面量、常量、`topicBuilder.xxx(tenantId)` → 前缀+`*` |
+
+无法解析时：展示表达式 + `<font color="comment">（destination 未解析）</font>`。
+
+### 2.4 读侧补强（类比 W06）
+
+| 中间件 | 补强来源 |
+|--------|----------|
+| RocketMQ | `RocketMQListener<T>`、`onMessage(T)` + `@RocketMQMessageListener` |
+| Kafka | `@KafkaListener` 方法参数类型；或方法内 `parseObject/parseArray(..., Xxx.class)`（与现有 W06 共享解析能力） |
+
+---
+
+## 3. 方案总览
+
+### 3.1 产品形态
+
+并入现有 `serialization-schema-checker`：
+
+- 同一 CLI / 流水线 / Schema Diff / 企微模板
+- 配置增加 `mq_patterns`（含 RocketMQ + Kafka）
+- 通知按 **Topic / destination** 分块；文案统一 `Topic -->`
+
+### 3.2 与现有链路
+
+```text
+Git Diff → 变更 Java 文件
+    ├─ Redis：W01~W05 + W06                    ← 已有
+    └─ MQ：RocketMQ（MQ01~）+ Kafka（MQ-K*）
+         + Listener / parse 补强（MQ-R）         ← 本方案
+              ↓
+        同一套 TypeSchema / SchemaDiffer / Skeleton / WeCom
+```
+
+对比区间：push **`before` → `after`**。
+
+### 3.3 核心原则
+
+1. 只关心**消息体对象 Schema**，不关心 Broker / ACL / 限流  
+2. **有结构变更即告警**；`block` 与缓存共用  
+3. **静态分析**；仅本仓 `src/main/java`  
+4. RocketMQ 与 Kafka **同一 Diff / 通知模型**，仅投递 AST 模式不同  
+
+---
+
+## 4. 检测模式设计
+
+### 4.1 生产侧 — RocketMQ
+
+| 模式 ID | 匹配表达式 | 提取 |
+|---------|------------|------|
+| MQ01 | `rocketMQTemplate.syncSend(dest, payload, …)` | dest、payload 类型 |
+| MQ02 | `asyncSend` / `syncSendOrderly` / `sendOneWay` 等 | 同上 |
+| MQ03 | `convertAndSend(dest, payload)` | 同上 |
+| MQ04 | `MessageBuilder.withPayload(obj)` 再 send | payload 类型 |
+| MQ05 | 先 `toJSONString`/`getObjectToString` 再 send String | unwrap 后类型 |
+
+### 4.2 生产侧 — Kafka
+
+| 模式 ID | 匹配表达式 | 提取 |
+|---------|------------|------|
+| MQ-K01 | `kafkaTemplate.send(topic, payload)` | topic、payload 类型 |
+| MQ-K02 | `kafkaTemplate.send(topic, key, payload)` | 同上（忽略分区 key） |
+| MQ-K03 | `send(ProducerRecord)` / `ListenableFuture` 封装若可解析 | topic + value 类型 |
+| MQ-K04 | 先 JSON 序列化为 String 再 `send(topic, json)` | unwrap 后类型 |
+
+Payload 为 `List<Xxx>` / `Collection` 时标记 **rootArray**，骨架为 JSON 数组（与 Redis List 一致）。
+
+**忽略**：
+
+- payload 为字面量、纯无结构 `String`/`byte[]`（无业务类型时）
+- 仅 Topic Admin API（`AdminClient.createTopics` 等）
+- destination 命中 `ignore.mq_destinations`
+
+### 4.3 消费侧辅助（不单独告警）
+
+| 模式 ID | 匹配 | 作用 |
+|---------|------|------|
+| MQ-R01 | `RocketMQListener<T>` / `@RocketMQMessageListener` | 补强同 destination 生产点 |
+| MQ-R02 | `@KafkaListener` + 参数类型 `T`（非 String） | 补强同 topic |
+| MQ-R03 | Listener 内 `parseObject`/`parseArray(..., Xxx.class)` | 补强（可复用 W06 检测器） |
+
+开关：`detection.mq_read_hints_enabled`（默认 true）。
+
+### 4.4 Schema Diff
+
+复用现有变更类型与注解规则。  
+序列化方言：首版按字段名；Jackson / Fastjson / Kafka JsonSerializer 差异必要时用 `manual_mappings`。
+
+---
+
+## 5. 报告与通知
+
+### 5.1 企微块（RocketMQ / Kafka 统一）
+
+```markdown
+- Topic --> `capital-topic:WALLET_DEDUCT`
+  > **通道**: `RocketMQ`
+  > **位置**: `WalletDeductProducer#send:38`
+  > **类型**: `WalletDeductReq`
+  > **value值由：** “{...}”
+  > **变更为：** “{...}”
+
+- Topic --> `patrol-store-food-safe:*`
+  > **通道**: `Kafka`
+  > **位置**: `PatrolServiceImpl#sendFoodSafeData:3140`
+  > **类型**: `List<CheckItemDetailVo>`
+  > **value值由：** “[{...}]”
+  > **变更为：** “[{...}]”
+```
+
+- 删除字段橙 `warning`；新增绿 `info`  
+- destination 未解析时灰色提示  
+- 可选后缀：「请评估消费积压与兼容反序列化」  
+
+「通道」字段用于区分中间件；若模板求简，可省略通道仅靠 Topic 形态区分。
+
+### 5.2 CI 控制台
+
+字段明细可标注 `RocketMQ` / `Kafka`；再输出与企微一致的 Markdown。
+
+---
+
+## 6. 配置草案
+
+```yaml
+detection:
+  patterns: [W01, W02, W03, W04, W05]
+  read_hints_enabled: true
+
+  mq_patterns:
+    # RocketMQ
+    - MQ01
+    - MQ02
+    - MQ03
+    - MQ04
+    - MQ05
+    # Kafka
+    - MQ-K01
+    - MQ-K02
+    - MQ-K03
+    - MQ-K04
+  mq_read_hints_enabled: true
+
+ignore:
+  mq_destinations:
+    - "*:TEST"
+    - "benchmark:*"
+
+manual_mappings:
+  - id: wallet-deduct-mq
+    writer_method: "jnpf.capital.module.wallet.mq.WalletDeductProducer#send"
+    key_pattern: "capital-topic:WALLET_DEDUCT"
+    value_type: "jnpf.model.capital.dto.WalletDeductReq"
+
+  - id: patrol-kafka-food-safe
+    writer_method: "jnpf.service.impl.PatrolServiceImpl#sendFoodSafeData"
+    key_pattern: "*-patrol-store-*"   # 按实际 topic 规则调整
+    value_type: "jnpf.model.analyses.CheckItemDetailVo"
+```
+
+---
+
+## 7. 分阶段交付
+
+### Phase M1 — MVP（RocketMQ + Kafka 基础投递）
+
+| 任务 | 说明 |
+|------|------|
+| MQ01/MQ02 | RocketMQ `syncSend` / `asyncSend` |
+| MQ-K01/MQ-K02 | Kafka `send(topic, payload)` / 三参 send |
+| destination 推断 | 字面量、常量、拼接；Kafka 动态 topic → `*` |
+| Schema Diff + 骨架通知 | 复用 ReportBuilder；可选「通道」行 |
+| 夹具 | `fixtures/mq/rocket-wallet/`、`fixtures/mq/kafka-patrol/` |
+| 配置 | `mq_patterns`（含 MQ-K*）、`ignore.mq_destinations` |
+
+**验收**：
+
+1. 删 `WalletDeductReq` 字段 → 企微出现 RocketMQ Topic 骨架变更  
+2. 删 `CheckItemDetailVo` 字段 → 企微出现 Kafka Topic 骨架变更  
+
+### Phase M2 — 增强
+
+| 任务 | 说明 |
+|------|------|
+| MQ03~MQ05、MQ-K03/K04 | convertAndSend、MessageBuilder、JSON 字符串发送、ProducerRecord |
+| MQ-R01~R03 | RocketMQ Listener + Kafka `@KafkaListener` / parse 补强 |
+| List 根数组骨架 | 巡店 `List<CheckItemDetailVo>` 等 |
+
+### Phase M3 — 运营
+
+| 任务 | 说明 |
+|------|------|
+| 积压风险提示文案 | 统一提示评估消费 lag / 积压 |
+| 更多夹具 | 值班 Kafka、中差评 Listener、IM Favorite 等 |
+| 分 webhook / 标题前缀 | 缓存 vs MQ 可选拆分 |
+
+---
+
+## 8. 风险与限制
+
+| 风险 | 缓解 |
+|------|------|
+| Kafka topic 按租户动态拼接 | 归一 `prefix:*`；`manual_mappings` |
+| RocketMQ / Kafka 混用同一 VO | 各投递点独立告警（符合预期） |
+| Listener 收 String、parse 在方法深处 | MQ-R03 + 复用 W06 AST |
+| 生产/消费跨模块对不齐 | 同仓索引 + destination 对齐；失败则仅写侧 |
+| Jackson / Fastjson / Kafka JsonSerializer 细节差 | 首版字段名；必要时方言或 mapping |
+| 只改消费未改生产类型 | 不告警（工具职责是消息体 Schema） |
+
+---
+
+## 9. 决策对齐
+
+| 项 | 结论 |
+|----|------|
+| 中间件范围 | **RocketMQ + Kafka**（本仓已确认）；Rabbit 暂不纳入 |
+| 对比区间 | `gitea.event.before` → `gitea.sha` |
+| 阻断 | 与现网 `mode` 共用 |
+| 级别 | 不引入 P0/P1/P2 产品展示 |
+| 交付 | 同一 jar；patterns 区分 Redis / MQ（含 MQ-K*） |
+
+---
+
+## 10. 下一步
+
+1. 评审本方案（RocketMQ + Kafka 模式表）  
+2. 按 **Phase M1** 开发（MQ01/02 + MQ-K01/K02）  
+3. 回写 `配置说明.md` / `CI集成说明.md` 正式配置项  
+4. 验收 Topic 建议：`capital-topic:WALLET_DEDUCT`（RocketMQ）、巡店/值班 Kafka topic  

docs/实施方案.md

@@ -1,10 +1,10 @@
-# 缓存序列化结构变更检测 — 实施方案
+# 序列化结构变更检测 — 实施方案
 
-> 版本：v0.2  
+> 版本：v0.3  
-> 日期：2026-07-14  
+> 日期：2026-07-15  
 > 技术栈：Java 11 + Maven + JavaParser  
-> 目标仓库：`redisCheck`（工具） / `jnpf-java-cloud`（被检测业务仓库）  
+> 目标仓库：`schemaCheck`（工具） / `jnpf-java-cloud`（被检测业务仓库）  
-> 当前阶段：**Phase 1 + Phase 2 已完成**，Phase 3 待做
+> 当前阶段：**Phase 1 + Phase 2 已完成**；Phase 3 运营待做；**MQ 扩展方案已文档落地**（见 `docs/MQ序列化结构检测方案.md`）
 
 ---
 
@@ -104,12 +104,12 @@
 schemaCheck 仓库
   ├── 开发 Java 分析工具
   ├── mvn package 打 fat-jar
-  ├── 发布到 Nexus：com.codechecker:cache-schema-checker:{version}
+  ├── 发布到 Nexus：com.codechecker:serialization-schema-checker:{version}
   └── 提供默认配置模板
 
 jnpf-java-cloud 仓库
-  ├── .gitea/workflows/cache-schema-check.yaml
+  ├── .gitea/workflows/serialization-schema-check.yaml
-  ├── .gitea/config/cache-schema-check-config.yaml
+  ├── .gitea/config/serialization-schema-check-config.yaml
   └── push 时下载 jar 并执行检测
 ```
 
@@ -119,11 +119,11 @@ jnpf-java-cloud 仓库
 flowchart TB
     subgraph Gitea["Gitea Push Pipeline"]
         A[push 事件] --> B[浅克隆 old/new 提交]
-        B --> C[下载 cache-schema-checker.jar]
+        B --> C[下载 serialization-schema-checker.jar]
         C --> D[java -jar 执行检测]
     end
 
-    subgraph Checker["cache-schema-checker (JDK 11)"]
+    subgraph Checker["serialization-schema-checker (JDK 11)"]
         D --> E[GitDiffScanner]
         E --> F[RedisWritePointDetector]
         F --> G[JavaSchemaExtractor]
@@ -143,7 +143,7 @@ flowchart TB
 1. **纯静态分析**：基于 Java 源码 AST + 符号解析，不启动 Spring 容器
 2. **Diff 驱动**：只分析本次 push 变更涉及的文件及其关联类型
 3. **本仓限定**：类型解析仅在业务仓库 `src/main/java` 范围内
-4. **可配置**：忽略规则、严重级别、通知开关、阻断开关均可 YAML 配置
+4. **可配置**：忽略规则、通知开关、阻断开关均可 YAML 配置
 5. **可演进**：已覆盖 JSON 字符串写入与 Template 直写 / Hash；后续可扩展读路径反向确认、报告落盘等
 
 ---
@@ -163,7 +163,7 @@ flowchart TB
 
 **不采用** Spoon / Eclipse JDT 的原因：JavaParser 足够覆盖第一版需求，依赖更轻，CLI 启动更快。
 
-**Lombok 处理策略**：第一版基于源码字段 + `@Data` 等注解推断序列化字段；对 `@Builder`、`@SuperBuilder` 等复杂场景标记为「低置信度」并降级为 P2 提示。后续可选集成 `lombok.ast` 或 delombok 预处理。
+**Lombok 处理策略**：基于源码字段 + `@Data` 等注解推断序列化字段；对 `@Builder`、`@SuperBuilder` 等复杂场景标记为低置信度提示。后续可选集成 `lombok.ast` 或 delombok 预处理。
 
 ---
 
@@ -175,7 +175,8 @@ schemaCheck/
 ├── docs/
 │   ├── 实施方案.md
 │   ├── 配置说明.md
-│   └── CI集成说明.md
+│   ├── CI集成说明.md
+│   └── MQ序列化结构检测方案.md   # MQ 消息体 Schema 扩展（方案）
 ├── src/
 │   ├── main/
 │   │   ├── resources/
@@ -195,8 +196,8 @@ schemaCheck/
 │       ├── resources/fixtures/{tenant,lock,template}/
 │       └── java/...
 ├── .gitea/
-│   ├── workflows/cache-schema-check.yaml
+│   ├── workflows/serialization-schema-check.yaml
-│   └── config/cache-schema-check-config.yaml
+│   └── config/serialization-schema-check-config.yaml
 └── target/                               # 构建产物
 ```
 
@@ -204,11 +205,11 @@ schemaCheck/
 
 ```xml
 <groupId>com.codechecker</groupId>
-<artifactId>cache-schema-checker</artifactId>
+<artifactId>serialization-schema-checker</artifactId>
 <version>1.0.0</version>
 ```
 
-打包为 **shaded/fat jar**，主类：`com.codechecker.cache.cli.CacheSchemaCheckerMain`
+打包为 **shaded/fat jar**，主类：`com.codechecker.cache.cli.SerializationSchemaCheckerMain`
 
 ---
 
@@ -217,8 +218,8 @@ schemaCheck/
 ### 6.1 CLI 参数
 
 ```bash
-java -jar cache-schema-checker-1.0.0.jar \
+java -jar serialization-schema-checker-1.0.0.jar \
-  --config .gitea/config/cache-schema-check-config.yaml \
+  --config .gitea/config/serialization-schema-check-config.yaml \
   --repo-root /path/to/jnpf-java-cloud \
   --old-sha abc123 \
   --new-sha def456 \
@@ -255,7 +256,7 @@ java -jar cache-schema-checker-1.0.0.jar \
 
 #### Step 1：加载配置
 
-读取 `cache-schema-check-config.yaml`，合并默认值（见 `docs/配置说明.md`）。
+读取 `serialization-schema-check-config.yaml`，合并默认值（见 `docs/配置说明.md`）。
 
 #### Step 2：Git Diff 扫描
 
@@ -282,9 +283,8 @@ git diff --name-only {old-sha} {new-sha} -- '*.java'
 | W01 | `redisUtil.insert(key, JSON.toJSONString(expr), ttl)` | key 表达式、value 表达式 | ✅ |
 | W02 | `redisTemplate.opsForValue().set(key, JSON.toJSONString(expr), ...)` | 同上 | ✅ |
 | W03 | `stringRedisTemplate.opsForValue().set(key, JsonUtil.getObjectToString(expr), ...)` | 同上 | ✅ |
-| W04 | `redisTemplate.opsForValue().set(key, expr, ...)` 且 expr 非字面量 | 直写对象类型 | ✅ Phase 2 |
+| W04 | `redisTemplate.opsForValue().set(key, expr, ...)` 且 expr 非字面量 | 直写对象类型 | ✅ |
-| W05 | `redisTemplate.opsForHash().put(key, field, expr)` | Hash 写出 value 类型 | ✅ Phase 2 |
+| W05 | `redisTemplate.opsForHash().put(key, field, expr)` | Hash 写出 value 类型 | ✅ |
-| W06 | `JSON.parseObject(cacheValue, Xxx.class)` | 辅助反向确认读取类型 | 未做 |
 
 **忽略规则**（自动）：
 
@@ -292,6 +292,19 @@ git diff --name-only {old-sha} {new-sha} -- '*.java'
 - 方法名含 `setIfAbsent`、`increment`、`delete`、`remove`、`expire` 等
 - key 匹配 `ignore.key_patterns` 配置（锁 / token / 登录计数等）
 
+#### Step 4b：读侧类型辅助（W06，非写入模式）
+
+W06 **不产生独立告警**，只扫描反序列化调用，用读到的 `Xxx.class` **补强**同文件（或同 key）写入点的 value 类型 / 根数组标记。
+
+| 匹配示例 | 作用 |
+|----------|------|
+| `JSON.parseObject(raw, Xxx.class)` | 补强对象类型 |
+| `JSON.parseArray(raw, Xxx.class)` / `JsonUtil.getJsonToList` | 补强 `List<Xxx>`（rootArray） |
+| `JsonUtil.getJsonToBean(raw, Xxx.class)` | 同上 |
+| 可关联到 `redisUtil.getString(key)` / `opsForValue().get(key)` | 同时补强 key 模式 |
+
+开关：`detection.read_hints_enabled`（默认 `true`）。优先级：`manual_mappings` > W06 补强 > 写侧 AST 推断。
+
 #### Step 5：类型推断
 
 对每个写入点的 `expr`，使用 JavaParser Symbol Solver 推断类型：
@@ -320,7 +333,7 @@ redisUtil.insert(buildCacheKey(encode), JSON.toJSONString(envelope), ttl);
 | `@JSONField(name = "xxx")` | 字段名映射 | ✅ |
 | `@JsonIgnore` | 排除字段 | ✅ |
 | `@JsonProperty("xxx")` | 字段名映射 | ✅ |
-| `@JsonIgnoreProperties({...})` | 类级忽略字段 | ✅ Phase 2 |
+| `@JsonIgnoreProperties({...})` | 类级忽略字段 | ✅ |
 | `@Schema` | 忽略（不影响序列化） | ✅ |
 
 #### Step 6：生成 JSON Schema
@@ -353,19 +366,19 @@ redisUtil.insert(buildCacheKey(encode), JSON.toJSONString(envelope), ttl);
 
 对比同一写入点在 old/new 两个版本的 `TypeSchema`，输出 `SchemaChange` 列表。
 
-**变更类型与严重级别**：
+**变更类型**（有结构差异即告警；`block` 下任意变更均阻断）：
 
-| 变更类型 | 示例 | 默认级别 |
+| 变更类型 | 示例 |
-|----------|------|----------|
+|----------|------|
-| `FIELD_REMOVED` | 删除 `dbName` | P0 |
+| `FIELD_REMOVED` | 删除 `dbName` |
-| `TYPE_CHANGED` | `linkList` 从数组变对象 | P0 |
+| `TYPE_CHANGED` | `linkList` 从数组变对象 |
-| `WRAPPER_ADDED` | 顶层增加 `vo` 包装 | P0 |
+| `WRAPPER_ADDED` | 顶层增加 `vo` 包装 |
-| `FIELD_PATH_MOVED` | `dbName` → `vo.dbName` | P0 |
+| `FIELD_PATH_MOVED` | `dbName` → `vo.dbName` |
-| `FIELD_ADDED` | 新增 `expiresAtMs` | P1 |
+| `FIELD_ADDED` | 新增 `expiresAtMs` |
-| `KEY_PATTERN_CHANGED` | key 常量变更 | P1 |
+| `KEY_PATTERN_CHANGED` | key 常量变更 |
-| `WRITE_POINT_REMOVED` | 删除缓存写入 | P1 |
+| `WRITE_POINT_REMOVED` | 删除缓存写入 |
-| `WRITE_POINT_ADDED` | 新增缓存写入 | P2 |
+| `WRITE_POINT_ADDED` | 新增缓存写入 |
-| `LOW_CONFIDENCE` | 类型推断失败 | P2 |
+| `LOW_CONFIDENCE` | 类型推断失败（仍提示，置信度较低） |
 
 #### Step 8：报告与通知
 
@@ -377,13 +390,13 @@ redisUtil.insert(buildCacheKey(encode), JSON.toJSONString(envelope), ttl);
 
 企微 Markdown 规则：
 
-- 抬头不含 mode / P0~P2 汇总；正文按 Key 展示骨架
+- 抬头不含 mode 汇总；正文按 Key 展示骨架
 - **删除字段**：旧骨架中橙色 `<font color="warning">`
 - **新增字段**：新骨架中绿色 `<font color="info">`
 - key 未解析时展示源码表达式 + 灰色「（key 未解析）」
 - 单条超 4096 UTF-8 字节时按 Key 拆成多条依次发送
 
-CI 控制台额外输出字段明细（含严重级别），再打印与企微一致的 Markdown。
+CI 控制台额外输出字段级变更明细（变更类型 + 位置 + 摘要），再打印与企微一致的 Markdown。
 
 调用企微 Webhook 发送 Markdown（支持 `--dry-run` 仅本地输出）。
 
@@ -430,16 +443,16 @@ manual_mappings:
 
 | 层级 | 位置 | 职责 |
 |------|------|------|
-| 默认配置 | 工具 jar 内 `default-config.yaml` | 检测模式、忽略规则、严重级别默认值 |
+| 默认配置 | 工具 jar 内 `default-config.yaml` | 检测模式、忽略规则等默认值 |
-| 业务覆盖 | `jnpf-java-cloud/.gitea/config/cache-schema-check-config.yaml` | mode、notify、include_modules、manual_mappings |
+| 业务覆盖 | `jnpf-java-cloud/.gitea/config/serialization-schema-check-config.yaml` | mode、notify、include_modules、manual_mappings |
 
 合并规则：**业务配置覆盖默认配置**，未声明的项沿用默认值。
 
 CLI 调用：
 
 ```bash
-java -jar cache-schema-checker.jar \
+java -jar serialization-schema-checker.jar \
-  --config .gitea/config/cache-schema-check-config.yaml \
+  --config .gitea/config/serialization-schema-check-config.yaml \
   ...
 ```
 
@@ -469,19 +482,19 @@ notify:
 详见 `docs/CI集成说明.md`。核心流程：
 
 ```yaml
-# jnpf-java-cloud/.gitea/workflows/cache-schema-check.yaml（要点）
+# jnpf-java-cloud/.gitea/workflows/serialization-schema-check.yaml（要点）
 # 检出：浅克隆 tip（depth 1）
 # 检测：--old-sha = gitea.event.before，--new-sha = gitea.sha
 # 按需 fetch before 提交对象，覆盖一次 push 的多 commit 累计 diff
 ```
 
-完整模板见 `docs/CI集成说明.md` / `.gitea/workflows/cache-schema-check.yaml`。
+完整模板见 `docs/CI集成说明.md` / `.gitea/workflows/serialization-schema-check.yaml`。
 
 ---
 
 ## 10. 分阶段交付计划
 
-### Phase 1 — MVP（约 1.5 周）✅
+### Phase 1 — MVP✅
 
 **目标**：跑通端到端链路，覆盖租户缓存典型场景。
 
@@ -491,10 +504,10 @@ notify:
 | Git diff 扫描 | 变更文件列表 | ✅ |
 | W01~W03 写入点检测 | 覆盖 JSON 字符串写入 | ✅ |
 | 基础 Schema 提取 | 支持普通类、内部类、List、嵌套 | ✅ |
-| Schema Diff P0/P1 | 字段增删、包装、路径迁移 | ✅ |
+| Schema Diff | 字段增删、包装、路径迁移 | ✅ |
 | 企微通知 | 按 Key 骨架 Markdown | ✅ |
 | notify/block / enabled | 配置驱动 | ✅ |
-| 夹具测试 | TenantVO/CacheEnvelope 样本 | ✅ |
+| 样本夹具测试 | TenantVO/CacheEnvelope 等 | ✅ |
 
 **验收标准**：
 
@@ -502,7 +515,7 @@ notify:
 - 流水线 push 后能收到企微通知
 - `mode=block` 时任意结构变更导致 exit 1
 
-### Phase 2 — 增强（约 1 周）✅
+### Phase 2 — 增强✅
 
 | 任务 | 说明 | 状态 |
 |------|------|------|
@@ -512,8 +525,9 @@ notify:
 | 忽略规则完善 | 锁/计数器/token/字面量/setIfAbsent | ✅ |
 | 多模块性能 | 并行读文件、索引批量装载、`manual_mappings` | ✅ |
 | 企微高亮 | 删除橙 `warning` / 新增绿 `info`；位置+类型通用项 | ✅ |
+| W06 读侧辅助 | `parseObject`/`parseArray` 等补强写入 value 类型 | ✅ |
 
-### Phase 3 — 运营（约 0.5 周）
+### Phase 3 — 运营
 
 | 任务 | 说明 |
 |------|------|
@@ -521,6 +535,23 @@ notify:
 | 误报反馈 | `suppressions` 按写入点 / change_types 精细忽略 |
 | 更多业务场景覆盖 | 考勤、文件下载进度等 |
 
+### Phase 4 — MQ 消息体结构检测（方案已落地，含 Kafka，开发待启）
+
+业务仓同时存在：
+
+- **RocketMQ**：`RocketMQTemplate.syncSend(topic:tag, dto)` / `RocketMQListener<T>`
+- **Kafka**：`KafkaTemplate.send(topic, vo|List)` / `@KafkaListener` + `parseObject`
+
+消息体字段变更会导致积压旧消息反序列化失败，风险模型与 Redis 同类。
+
+| 任务 | 说明 | 状态 |
+|------|------|------|
+| 方案文档 | RocketMQ + Kafka 模式、destination、复用 Schema Diff/企微 | ✅ 见专用文档 |
+| Phase M1 | RocketMQ syncSend + Kafka send + 骨架通知 + 夹具 | 待启 |
+| Phase M2/M3 | Listener/parse 补强、更多投递形态、运营 | 待启 |
+
+**专用方案：** [`docs/MQ序列化结构检测方案.md`](./MQ序列化结构检测方案.md)
+
 ---
 
 ## 11. 测试策略
@@ -532,12 +563,14 @@ notify:
 - `RedisWritePointDetectorTest`：各种写入 AST 模式匹配
 - `RedisKeyResolverTest`：常量、format、拼接推断
 
-### 11.2 夹具集成测试
+### 11.2 样本夹具测试（fixtures）
 
-在 `src/test/resources/fixtures/` 放置真实业务代码片段（从 `jnpf-java-cloud` 提取并脱敏），模拟 old/new 两个版本：
+「夹具」= 放在测试资源里的**脱敏源码样本**（不是连真实 Redis / 不是起 Gitea 流水线）。
 
-| 夹具 | 验证点 |
+路径：`src/test/resources/fixtures/`。测试代码加载这些 `.txt`/Java 片段，在内存中跑检测器 / Schema 对比，用来验证典型业务场景是否被正确识别。
-|------|--------|
+
+| 夹具目录 | 验证点 |
+|----------|--------|
 | `fixtures/tenant/` | 包装结构变更（TenantVO → CacheEnvelope） |
 | `fixtures/lock/` | 锁/计数器/token 应被忽略 |
 | `fixtures/template/` | W04 Template 直写 |
@@ -567,12 +600,20 @@ notify:
 | # | 决策项 | 结论 |
 |---|--------|------|
 | 1 | 阻断范围 | `block` 模式下 **任意结构变更均阻断**（exit 1） |
-| 2 | 发布坐标 | 独立产物 `com.codechecker:cache-schema-checker:1.0.0` |
+| 2 | 发布坐标 | 独立产物 `com.codechecker:serialization-schema-checker:1.0.0` |
 | 3 | 配置归属 | **双层配置**：jar 内 `default-config.yaml` + 业务仓覆盖合并 |
-| 4 | 上线策略 | 先 `notify` 观察 **1 周**，稳定后手动切 `block` |
+| 4 | 上线策略 | 先 `notify` ，稳定后手动切 `block` |
 | 5 | 检测范围 | **仅 `src/main/java`**，不扫描测试代码 |
 
-以上决策已纳入实施方案；**Phase 1 / Phase 2 已交付**，可进入 Phase 3 或业务仓全量观察。
+以上决策已纳入实施方案；**Phase 1 / Phase 2 已交付**。缓存侧可进入 Phase 3；MQ 侧以 [`MQ序列化结构检测方案.md`](./MQ序列化结构检测方案.md) 为准评审后开发。
+
+相关文档：
+
+| 文档 | 内容 |
+|------|------|
+| `docs/配置说明.md` | 缓存检测双层配置 |
+| `docs/CI集成说明.md` | 流水线 before/after、排障 |
+| `docs/MQ序列化结构检测方案.md` | MQ 消息体 Schema 监控方案（扩展） |
 
 ---
 
@@ -598,7 +639,6 @@ public class WritePoint {
 
 ```java
 public class SchemaChange {
-    Severity severity;          // P0, P1, P2
     ChangeType changeType;      // FIELD_REMOVED, WRAPPER_ADDED, ...
     String keyPattern;
     String writeLocation;       // class#method:line

docs/配置说明.md

@@ -1,8 +1,8 @@
-# 缓存序列化结构检测 — 配置说明
+# 序列化结构检测 — 配置说明
 
-> **双层配置**：工具 jar 内置 `default-config.yaml`（默认） + 业务仓库 `.gitea/config/cache-schema-check-config.yaml`（覆盖）  
+> **双层配置**：工具 jar 内置 `default-config.yaml`（默认） + 业务仓库 `.gitea/config/serialization-schema-check-config.yaml`（覆盖）  
-> 工具坐标：`com.codechecker:cache-schema-checker:1.0.0`  
+> 工具坐标：`com.codechecker:serialization-schema-checker:1.0.0`  
-> 主类：`com.codechecker.cache.cli.CacheSchemaCheckerMain`
+> 主类：`com.codechecker.cache.cli.SerializationSchemaCheckerMain`
 
 ---
 
@@ -11,7 +11,7 @@
 ```text
 jar 内 default-config.yaml（工具仓维护）
         ↓ 深度合并
-业务仓 cache-schema-check-config.yaml（业务仓维护）
+业务仓 serialization-schema-check-config.yaml（业务仓维护）
         ↓
 最终生效配置
 ```
@@ -23,7 +23,7 @@ jar 内 default-config.yaml（工具仓维护）
 ### 1.1 业务仓最小配置示例
 
 ```yaml
-# jnpf-java-cloud/.gitea/config/cache-schema-check-config.yaml
+# jnpf-java-cloud/.gitea/config/serialization-schema-check-config.yaml
 enabled: true
 mode: notify
 
@@ -43,6 +43,7 @@ include_modules:
 由 `redisCheck` 仓库维护，随 jar 发布，默认包含：
 
 - `detection.patterns`：**W01~W05**（JSON 字符串写入 + Template 直写 + Hash）
+- `detection.read_hints_enabled`：W06 读侧反序列化类型辅助（默认 true）
 - `ignore.key_patterns`（锁 / 计数器 / token）
 - `detection.min_confidence`、`max_field_depth`
 - `mode: notify`、`enabled: true`
@@ -77,7 +78,7 @@ notify:
   # 无变更时是否也发通知（一般 false）
   notify_on_clean: false
   # 消息标题前缀
-  title_prefix: "[缓存结构变更]"
+  title_prefix: "[序列化结构变更]"
 
 # 忽略规则
 ignore:
@@ -106,17 +107,15 @@ detection:
     - W04   # redisTemplate 直写对象
     - W05   # opsForHash().put
 
-  # 类型推断最低置信度，低于此值仅输出 P2 提示
+  # W06：读侧反序列化类型辅助（不产生独立告警）
+  read_hints_enabled: true
+
+  # 类型推断最低置信度，低于此值标记为低置信度提示
   min_confidence: 0.6
 
   # 字段展开最大深度（防止循环引用死循环）
   max_field_depth: 8
 
-# 严重级别覆盖（可选）
-severity_overrides:
-  FIELD_ADDED: P1
-  WRITE_POINT_ADDED: P2
-
 # 人工补充映射（自动推断失败或需精确指定时使用）
 manual_mappings:
   - id: tenant-db-content
@@ -183,7 +182,7 @@ mode: block
 | `webhook_url` | string | `""` | 企微机器人 Webhook 完整 URL（优先） |
 | `webhook_env` | string | — | 兼容旧字段；值为 `http` 开头时当作 URL |
 | `notify_on_clean` | boolean | false | 无变更时是否通知 |
-| `title_prefix` | string | [缓存结构变更] | 消息标题前缀 |
+| `title_prefix` | string | [序列化结构变更] | 消息标题前缀 |
 
 ### 3.4 ignore.key_patterns
 
@@ -216,6 +215,23 @@ detection:
   patterns: [W01, W02, W03]
 ```
 
+### 3.5.1 detection.read_hints_enabled（W06 辅助）
+
+| 字段 | 类型 | 默认 | 说明 |
+|------|------|------|------|
+| `read_hints_enabled` | boolean | true | 扫描 `parseObject` / `parseArray` / `getJsonToBean` 等，补强同文件或同 key 写入点的 value 类型 |
+
+- **不是写入模式**：不会单独因为「多了一处 parse」而告警  
+- 能关联到 `redis get(key)` 时，还可补强 unresolved key  
+- 覆盖优先级：`manual_mappings` > W06 > 写侧 AST
+
+关闭示例：
+
+```yaml
+detection:
+  read_hints_enabled: false
+```
+
 ### 3.6 manual_mappings
 
 当自动推断不准确时使用。匹配优先级 **高于** 自动推断（按 `类全名#方法名` 覆盖 key 模式与 value 类型）。
@@ -266,10 +282,10 @@ suppressions:
 
 ### 5.1 结构说明
 
-- 抬头：仓库、分支、提交、提交人、时间（**不再**展示 mode / P0P1P2 汇总）
+- 抬头：仓库、分支、提交、提交人、时间（不展示 mode）
 - 正文：按 **一个 Redis Key 一块**，展示位置、类型与前后序列化骨架
 - 超长（UTF-8 > 4096 字节）时按 key **拆成多条**消息依次发送
-- CI 控制台另打「字段明细」（含 P0/P1/P2），企微侧不分级别
+- CI 控制台另打「字段明细」（变更类型 / 位置 / 摘要），企微侧按骨架展示
 
 ### 5.2 字段高亮颜色
 
@@ -283,7 +299,7 @@ suppressions:
 ### 5.3 示例（已解析 key）
 
 ```markdown
-## [缓存结构变更] jnpf-java-cloud
+## [序列化结构变更] jnpf-java-cloud
 
 > **分支**: code/redis_change_detection_v1.0
 > **提交**: cedd161c → 67c8a6eb
@@ -315,8 +331,6 @@ suppressions:
 
 ## 6. 推荐上线配置
 
-### 6.1 观察期（第 1 周）
-
 业务仓默认配置：
 
 ```yaml
@@ -327,23 +341,15 @@ notify:
   enabled: true
   webhook_url: ""   # 由流水线写入真实 Webhook
 
-include_modules:
+include_modules: []  # 全仓
-  - jnpf-tenant
 ```
 
-观察满 1 周、确认误报可接受后，手动切换：
+---
 
-```yaml
+## 7. 扩展：MQ 消息体检测（方案阶段）
-mode: block
-include_modules: []  # 扩至全仓
-```
 
-### 6.2 全量启用（观察期结束后）
+MQ（**RocketMQ + Kafka**）消息体 Schema 变更监控方案已单独成文，**实现前不要求业务仓增配**。
 
-```yaml
+详见：[MQ序列化结构检测方案.md](./MQ序列化结构检测方案.md)
-enabled: true
+
-mode: block
+届时预计新增：`detection.mq_patterns`（含 `MQ01~` 与 `MQ-K01~`）、`detection.mq_read_hints_enabled`、`ignore.mq_destinations`。
-include_modules: []  # 空表示全部模块
-detection:
-  patterns: [W01, W02, W03, W04, W05]
-```

pom.xml

@@ -5,12 +5,12 @@
     <modelVersion>4.0.0</modelVersion>
 
     <groupId>com.codechecker</groupId>
-    <artifactId>cache-schema-checker</artifactId>
+    <artifactId>serialization-schema-checker</artifactId>
     <version>1.0.0</version>
     <packaging>jar</packaging>
 
-    <name>cache-schema-checker</name>
+    <name>serialization-schema-checker</name>
-    <description>基于 JavaParser 的缓存 value 序列化结构变更检测器</description>
+    <description>基于 JavaParser 的缓存/MQ value 序列化结构变更检测器</description>
 
     <properties>
         <maven.compiler.source>11</maven.compiler.source>
@@ -74,7 +74,7 @@
     </dependencies>
 
     <build>
-        <finalName>cache-schema-checker-${project.version}</finalName>
+        <finalName>serialization-schema-checker-${project.version}</finalName>
         <plugins>
             <plugin>
                 <groupId>org.apache.maven.plugins</groupId>
@@ -101,7 +101,7 @@
                             <transformers>
                                 <transformer
                                         implementation="org.apache.maven.plugins.shade.resource.ManifestResourceTransformer">
-                                    <mainClass>com.codechecker.cache.cli.CacheSchemaCheckerMain</mainClass>
+                                    <mainClass>com.codechecker.cache.cli.SerializationSchemaCheckerMain</mainClass>
                                 </transformer>
                                 <transformer
                                         implementation="org.apache.maven.plugins.shade.resource.ServicesResourceTransformer"/>

src/main/java/com/codechecker/cache/analyze/SchemaCheckAnalyzer.java

@@ -1,6 +1,8 @@
 package com.codechecker.cache.analyze;
 
 import com.codechecker.cache.config.CheckerConfig;
+import com.codechecker.cache.detector.CacheReadHint;
+import com.codechecker.cache.detector.CacheReadHintDetector;
 import com.codechecker.cache.detector.RedisWritePointDetector;
 import com.codechecker.cache.detector.WritePoint;
 import com.codechecker.cache.diff.ChangeType;
@@ -111,6 +113,10 @@ public class SchemaCheckAnalyzer {
             List<WritePoint> newWps = detectorNew.detect(path, newContent);
             List<WritePoint> oldWps = oldContent == null
                     ? new ArrayList<>() : detectorOld.detect(path, oldContent);
+            applyReadHints(newWps, path, newContent, newIndex);
+            if (oldContent != null) {
+                applyReadHints(oldWps, path, oldContent, oldIndex);
+            }
             newWps.forEach(this::applyManualMappings);
             oldWps.forEach(this::applyManualMappings);
 
@@ -268,8 +274,10 @@ public class SchemaCheckAnalyzer {
                 paths.add(c.getOldValue());
             }
             if (c.getChangeType() == ChangeType.WRAPPER_ADDED
-                    && c.getFieldPath() != null) {
+                    || c.getChangeType() == ChangeType.WRAPPER_REMOVED) {
-                paths.add(c.getFieldPath());
+                if (c.getFieldPath() != null) {
+                    paths.add(c.getFieldPath());
+                }
             }
         }
         return paths;
@@ -425,6 +433,92 @@ public class SchemaCheckAnalyzer {
         }
     }
 
+    /**
+     * W06：用同文件读侧反序列化类型提示，补强低置信度 / 缺类型的写入点。
+     * manual_mappings 仍在其后执行，可覆盖本补强结果。
+     */
+    private void applyReadHints(List<WritePoint> writePoints, String path, String content,
+                                SourceIndex index) {
+        if (!config.getDetection().isReadHintsEnabled()
+                || writePoints == null || writePoints.isEmpty()) {
+            return;
+        }
+        List<CacheReadHint> hints = new CacheReadHintDetector(index).detect(path, content);
+        if (hints.isEmpty()) {
+            return;
+        }
+        for (WritePoint wp : writePoints) {
+            enrichWritePointFromHints(wp, hints);
+        }
+    }
+
+    private void enrichWritePointFromHints(WritePoint wp, List<CacheReadHint> hints) {
+        boolean needType = wp.getResolvedValueType() == null || wp.getResolvedValueType().isEmpty()
+                || wp.getConfidence() < config.getDetection().getMinConfidence();
+        boolean needKey = isUnresolvedKey(wp.getResolvedKeyPattern());
+        if (!needType && !needKey) {
+            return;
+        }
+
+        CacheReadHint best = null;
+        int bestScore = -1;
+        for (CacheReadHint hint : hints) {
+            int score = scoreHint(wp, hint);
+            if (score > bestScore) {
+                bestScore = score;
+                best = hint;
+            }
+        }
+        if (best == null || bestScore <= 0) {
+            return;
+        }
+        if (needType && best.getResolvedValueType() != null) {
+            wp.setResolvedValueType(best.getResolvedValueType());
+            wp.setRootArray(best.isRootArray());
+            wp.setConfidence(Math.max(wp.getConfidence(), best.getConfidence()));
+        }
+        if (needKey && best.getResolvedKeyPattern() != null
+                && !isUnresolvedKey(best.getResolvedKeyPattern())) {
+            wp.setResolvedKeyPattern(best.getResolvedKeyPattern());
+            if (best.getKeyExpression() != null) {
+                // 保留写入侧原始表达式，仅补 key 模式
+            }
+        }
+    }
+
+    /** 匹配得分：同 key > 同类同方法 > 同类；无交集则 0。 */
+    private int scoreHint(WritePoint wp, CacheReadHint hint) {
+        if (hint.getResolvedValueType() == null) {
+            return 0;
+        }
+        boolean sameClass = wp.getEnclosingClass() != null
+                && wp.getEnclosingClass().equals(hint.getEnclosingClass());
+        if (!sameClass) {
+            // 跨文件仅允许 key 模式已解析且一致
+            if (wp.getResolvedKeyPattern() != null
+                    && wp.getResolvedKeyPattern().equals(hint.getResolvedKeyPattern())
+                    && !isUnresolvedKey(wp.getResolvedKeyPattern())) {
+                return 40;
+            }
+            return 0;
+        }
+        int score = 10;
+        if (wp.getEnclosingMethod() != null
+                && wp.getEnclosingMethod().equals(hint.getEnclosingMethod())) {
+            score += 20;
+        }
+        if (wp.getResolvedKeyPattern() != null
+                && wp.getResolvedKeyPattern().equals(hint.getResolvedKeyPattern())
+                && !isUnresolvedKey(wp.getResolvedKeyPattern())) {
+            score += 50;
+        } else if (hint.getResolvedKeyPattern() != null
+                && !isUnresolvedKey(hint.getResolvedKeyPattern())
+                && isUnresolvedKey(wp.getResolvedKeyPattern())) {
+            score += 30;
+        }
+        return score;
+    }
+
     private void collectTypeNames(String content, Set<String> out) {
         if (content == null || content.isEmpty()) {
             return;

src/main/java/com/codechecker/cache/cli/SerializationSchemaCheckerMain.java

@@ -17,11 +17,11 @@ import java.util.concurrent.Callable;
 /**
  * 命令行入口。退出码：0 通过 / 1 阻断 / 2 执行错误。
  */
-@Command(name = "cache-schema-checker",
+@Command(name = "serialization-schema-checker",
         mixinStandardHelpOptions = true,
-        version = "cache-schema-checker 1.0.0",
+        version = "serialization-schema-checker 1.0.0",
-        description = "检测两次提交间缓存 value 序列化结构变更并通过企微机器人通知。")
+        description = "检测两次提交间缓存/MQ 等 value 序列化结构变更并通过企微机器人通知。")
-public class CacheSchemaCheckerMain implements Callable<Integer> {
+public class SerializationSchemaCheckerMain implements Callable<Integer> {
 
     @Option(names = "--config", required = true, description = "业务仓库检测配置文件路径")
     private Path configPath;
@@ -56,12 +56,12 @@ public class CacheSchemaCheckerMain implements Callable<Integer> {
             CheckerConfig config = ConfigLoader.load(configPath);
 
             if (!config.isEnabled()) {
-                System.out.println("[cache-schema-checker] 总开关 enabled=false，跳过检测。");
+                System.out.println("[serialization-schema-checker] 总开关 enabled=false，跳过检测。");
                 return 0;
             }
 
             if (oldSha == null || oldSha.trim().isEmpty()) {
-                System.out.println("[cache-schema-checker] 无对比基准提交，跳过检测。");
+                System.out.println("[serialization-schema-checker] 无对比基准提交，跳过检测。");
                 return 0;
             }
 
@@ -75,7 +75,6 @@ public class CacheSchemaCheckerMain implements Callable<Integer> {
             report.setRepository(repository != null ? repository : root.getFileName().toString());
 
             ReportBuilder builder = new ReportBuilder(config.getNotify().getTitlePrefix());
-            // CI：字段明细 + 完整企微 Markdown
             System.out.println(builder.toConsole(report));
 
             boolean shouldNotify = config.getNotify().isEnabled()
@@ -84,24 +83,24 @@ public class CacheSchemaCheckerMain implements Callable<Integer> {
                 String webhook = config.getNotify().getWebhookUrl();
                 List<String> messages = builder.toWeComMessages(report);
                 int ok = new WeComNotifier().sendMarkdownMessages(webhook, messages);
-                System.out.println("[cache-schema-checker] 企微通知发送: "
+                System.out.println("[serialization-schema-checker] 企微通知发送: "
                         + ok + "/" + messages.size()
                         + (messages.size() > 1 ? "（已按 key 拆分）" : ""));
             }
 
             if (report.isBlocked()) {
-                System.out.println("[cache-schema-checker] block 模式命中，流水线将被阻断（exit 1）。");
+                System.out.println("[serialization-schema-checker] block 模式命中，流水线将被阻断（exit 1）。");
             }
             return report.getExitCode();
         } catch (Exception e) {
-            System.err.println("[cache-schema-checker] 执行错误: " + e.getMessage());
+            System.err.println("[serialization-schema-checker] 执行错误: " + e.getMessage());
             e.printStackTrace();
             return 2;
         }
     }
 
     public static void main(String[] args) {
-        int exitCode = new CommandLine(new CacheSchemaCheckerMain()).execute(args);
+        int exitCode = new CommandLine(new SerializationSchemaCheckerMain()).execute(args);
         System.exit(exitCode);
     }
 }

src/main/java/com/codechecker/cache/config/CheckerConfig.java

@@ -41,7 +41,7 @@ public class CheckerConfig {
         /** 企微机器人 Webhook 完整 URL */
         private String webhookUrl = "";
         private boolean notifyOnClean = false;
-        private String titlePrefix = "[缓存结构变更]";
+        private String titlePrefix = "[序列化结构变更]";
 
         public boolean isEnabled() {
             return enabled;
@@ -110,6 +110,8 @@ public class CheckerConfig {
         private List<String> patterns = new ArrayList<>();
         private double minConfidence = 0.6;
         private int maxFieldDepth = 8;
+        /** W06：是否启用读侧反序列化类型辅助补强 */
+        private boolean readHintsEnabled = true;
 
         public List<String> getPatterns() {
             return patterns;
@@ -134,6 +136,14 @@ public class CheckerConfig {
         public void setMaxFieldDepth(int maxFieldDepth) {
             this.maxFieldDepth = maxFieldDepth;
         }
+
+        public boolean isReadHintsEnabled() {
+            return readHintsEnabled;
+        }
+
+        public void setReadHintsEnabled(boolean readHintsEnabled) {
+            this.readHintsEnabled = readHintsEnabled;
+        }
     }
 
     public static class ManualMapping {

src/main/java/com/codechecker/cache/config/ConfigLoader.java

@@ -87,7 +87,7 @@ public final class ConfigLoader {
         n.setEnabled(bool(notify, "enabled", true));
         n.setWebhookUrl(resolveWebhookUrl(notify));
         n.setNotifyOnClean(bool(notify, "notify_on_clean", false));
-        n.setTitlePrefix(str(notify, "title_prefix", "[缓存结构变更]"));
+        n.setTitlePrefix(str(notify, "title_prefix", "[序列化结构变更]"));
 
         Map<String, Object> ignore = asMap(map.get("ignore"));
         CheckerConfig.Ignore ig = config.getIgnore();
@@ -100,6 +100,7 @@ public final class ConfigLoader {
         d.setPatterns(strList(detection.get("patterns")));
         d.setMinConfidence(dbl(detection, "min_confidence", 0.6));
         d.setMaxFieldDepth((int) lng(detection, "max_field_depth", 8));
+        d.setReadHintsEnabled(bool(detection, "read_hints_enabled", true));
 
         Map<String, Object> severityOverrides = asMap(map.get("severity_overrides"));
         Map<String, String> so = new LinkedHashMap<>();

src/main/java/com/codechecker/cache/detector/CacheReadHint.java

@@ -0,0 +1,92 @@
+package com.codechecker.cache.detector;
+
+/**
+ * 读侧反序列化类型提示（W06）：从 parseObject / getJsonToBean 等推断「缓存 value 被当成什么类型用」。
+ * 不单独产生告警，仅用于补强同文件/同 key 写入点的 value 类型。
+ */
+public class CacheReadHint {
+
+    private String filePath;
+    private int lineNumber;
+    private String enclosingClass;
+    private String enclosingMethod;
+    /** 推断出的 key 模式；无法关联 redis get 时为 null */
+    private String resolvedKeyPattern;
+    private String keyExpression;
+    /** value 元素/对象 FQN */
+    private String resolvedValueType;
+    private boolean rootArray;
+    private double confidence = 0.7;
+
+    public String getFilePath() {
+        return filePath;
+    }
+
+    public void setFilePath(String filePath) {
+        this.filePath = filePath;
+    }
+
+    public int getLineNumber() {
+        return lineNumber;
+    }
+
+    public void setLineNumber(int lineNumber) {
+        this.lineNumber = lineNumber;
+    }
+
+    public String getEnclosingClass() {
+        return enclosingClass;
+    }
+
+    public void setEnclosingClass(String enclosingClass) {
+        this.enclosingClass = enclosingClass;
+    }
+
+    public String getEnclosingMethod() {
+        return enclosingMethod;
+    }
+
+    public void setEnclosingMethod(String enclosingMethod) {
+        this.enclosingMethod = enclosingMethod;
+    }
+
+    public String getResolvedKeyPattern() {
+        return resolvedKeyPattern;
+    }
+
+    public void setResolvedKeyPattern(String resolvedKeyPattern) {
+        this.resolvedKeyPattern = resolvedKeyPattern;
+    }
+
+    public String getKeyExpression() {
+        return keyExpression;
+    }
+
+    public void setKeyExpression(String keyExpression) {
+        this.keyExpression = keyExpression;
+    }
+
+    public String getResolvedValueType() {
+        return resolvedValueType;
+    }
+
+    public void setResolvedValueType(String resolvedValueType) {
+        this.resolvedValueType = resolvedValueType;
+    }
+
+    public boolean isRootArray() {
+        return rootArray;
+    }
+
+    public void setRootArray(boolean rootArray) {
+        this.rootArray = rootArray;
+    }
+
+    public double getConfidence() {
+        return confidence;
+    }
+
+    public void setConfidence(double confidence) {
+        this.confidence = confidence;
+    }
+}

src/main/java/com/codechecker/cache/detector/CacheReadHintDetector.java

@@ -0,0 +1,189 @@
+package com.codechecker.cache.detector;
+
+import com.codechecker.cache.key.RedisKeyResolver;
+import com.codechecker.cache.schema.SourceIndex;
+import com.github.javaparser.StaticJavaParser;
+import com.github.javaparser.ast.CompilationUnit;
+import com.github.javaparser.ast.body.CallableDeclaration;
+import com.github.javaparser.ast.body.ClassOrInterfaceDeclaration;
+import com.github.javaparser.ast.body.VariableDeclarator;
+import com.github.javaparser.ast.expr.ClassExpr;
+import com.github.javaparser.ast.expr.Expression;
+import com.github.javaparser.ast.expr.MethodCallExpr;
+import com.github.javaparser.ast.expr.NameExpr;
+import com.github.javaparser.ast.type.ClassOrInterfaceType;
+import com.github.javaparser.ast.type.Type;
+
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.HashSet;
+import java.util.List;
+import java.util.Locale;
+import java.util.Optional;
+import java.util.Set;
+
+/**
+ * W06：扫描读侧反序列化，提取「缓存字符串 → 业务类型」提示，供写入点类型补强。
+ * <p>典型模式：</p>
+ * <pre>
+ * String raw = redisUtil.getString(key);
+ * Foo vo = JSON.parseObject(raw, Foo.class);
+ * List&lt;Foo&gt; list = JSON.parseArray(raw, Foo.class);
+ * </pre>
+ */
+public class CacheReadHintDetector {
+
+    private static final Set<String> OBJECT_PARSE = new HashSet<>(Arrays.asList(
+            "parseObject", "parse", "getJsonToBean", "toJavaObject", "readValue"));
+    private static final Set<String> ARRAY_PARSE = new HashSet<>(Arrays.asList(
+            "parseArray", "getJsonToList", "parseArrayObject"));
+    private static final Set<String> REDIS_GET = new HashSet<>(Arrays.asList(
+            "get", "getString", "opsForValue"));
+
+    private final SourceIndex index;
+    private final RedisKeyResolver keyResolver;
+
+    public CacheReadHintDetector(SourceIndex index) {
+        this.index = index;
+        this.keyResolver = new RedisKeyResolver(index);
+    }
+
+    public List<CacheReadHint> detect(String filePath, String content) {
+        List<CacheReadHint> result = new ArrayList<>();
+        if (content == null || content.isEmpty()) {
+            return result;
+        }
+        CompilationUnit cu;
+        try {
+            cu = StaticJavaParser.parse(content);
+        } catch (RuntimeException e) {
+            return result;
+        }
+
+        for (MethodCallExpr mce : cu.findAll(MethodCallExpr.class)) {
+            CacheReadHint hint = tryParseHint(mce, filePath);
+            if (hint != null) {
+                result.add(hint);
+            }
+        }
+        return result;
+    }
+
+    private CacheReadHint tryParseHint(MethodCallExpr mce, String filePath) {
+        String name = mce.getNameAsString();
+        boolean array = ARRAY_PARSE.contains(name);
+        boolean object = OBJECT_PARSE.contains(name);
+        if (!array && !object) {
+            return null;
+        }
+        if (mce.getArguments().size() < 2) {
+            return null;
+        }
+
+        Expression classArg = mce.getArgument(1);
+        // readValue(str, TypeReference) 等暂不支持；要求 ClassLiteral Xxx.class
+        if (!(classArg instanceof ClassExpr)) {
+            // 部分 API：parseObject(str, Xxx.class, Feature...) 仍是第 2 参
+            return null;
+        }
+        Type type = ((ClassExpr) classArg).getType();
+        if (!(type instanceof ClassOrInterfaceType)) {
+            return null;
+        }
+
+        ClassOrInterfaceDeclaration enclosing = mce
+                .findAncestor(ClassOrInterfaceDeclaration.class).orElse(null);
+        String enclosingFqn = enclosing == null
+                ? "<unknown>"
+                : enclosing.getFullyQualifiedName().orElse(enclosing.getNameAsString());
+        SourceIndex.IndexedType context = index.get(enclosingFqn);
+
+        String fqn = index.resolveFqn(((ClassOrInterfaceType) type).getNameWithScope(), context);
+        if (fqn == null) {
+            fqn = index.resolveFqn(((ClassOrInterfaceType) type).getNameAsString(), context);
+        }
+        if (fqn == null) {
+            return null;
+        }
+
+        CacheReadHint hint = new CacheReadHint();
+        hint.setFilePath(filePath);
+        hint.setLineNumber(mce.getBegin().map(p -> p.line).orElse(0));
+        hint.setEnclosingClass(enclosingFqn);
+        hint.setEnclosingMethod(mce.findAncestor(CallableDeclaration.class)
+                .map(CallableDeclaration::getNameAsString).orElse("<unknown>"));
+        hint.setResolvedValueType(fqn);
+        hint.setRootArray(array || "parseArray".equals(name) || "getJsonToList".equals(name));
+        hint.setConfidence(0.7);
+
+        Expression rawExpr = mce.getArgument(0);
+        Optional<RedisGetRef> getRef = findRedisGetForVar(rawExpr, mce);
+        if (getRef.isPresent()) {
+            hint.setKeyExpression(getRef.get().keyExpr.toString());
+            hint.setResolvedKeyPattern(keyResolver.resolve(
+                    getRef.get().keyExpr, enclosing, context));
+            hint.setConfidence(0.85);
+        }
+        return hint;
+    }
+
+    /**
+     * 若 parse 的第 1 参是局部变量，追溯其是否来自 redis get(key)。
+     */
+    private Optional<RedisGetRef> findRedisGetForVar(Expression rawExpr, MethodCallExpr parseCall) {
+        if (!(rawExpr instanceof NameExpr)) {
+            // 直接 parseObject(redis.get(key), Xxx.class)
+            if (rawExpr instanceof MethodCallExpr) {
+                return extractGetKey((MethodCallExpr) rawExpr);
+            }
+            return Optional.empty();
+        }
+        String varName = ((NameExpr) rawExpr).getNameAsString();
+        Optional<CallableDeclaration> callable = parseCall.findAncestor(CallableDeclaration.class);
+        if (!callable.isPresent()) {
+            return Optional.empty();
+        }
+        for (VariableDeclarator var : callable.get().findAll(VariableDeclarator.class)) {
+            if (!var.getNameAsString().equals(varName) || !var.getInitializer().isPresent()) {
+                continue;
+            }
+            Expression init = var.getInitializer().get();
+            if (init instanceof MethodCallExpr) {
+                return extractGetKey((MethodCallExpr) init);
+            }
+        }
+        return Optional.empty();
+    }
+
+    private Optional<RedisGetRef> extractGetKey(MethodCallExpr call) {
+        String name = call.getNameAsString();
+        String scope = call.getScope().map(Expression::toString).orElse("").toLowerCase(Locale.ROOT);
+        boolean redisScope = scope.contains("redis") || scope.contains("opsforvalue")
+                || scope.contains("boundvalueops");
+        if ("get".equals(name) || "getString".equals(name)) {
+            if (!redisScope && !REDIS_GET.contains(name)) {
+                // getString 也常见于 RedisUtil
+                if (!"getString".equals(name)) {
+                    return Optional.empty();
+                }
+            }
+            if (call.getArguments().isEmpty()) {
+                return Optional.empty();
+            }
+            return Optional.of(new RedisGetRef(call.getArgument(0)));
+        }
+        // redisTemplate.opsForValue().get(key)
+        if ("get".equals(name) && scope.contains("opsforvalue") && !call.getArguments().isEmpty()) {
+            return Optional.of(new RedisGetRef(call.getArgument(0)));
+        }
+        return Optional.empty();
+    }
+
+    private static final class RedisGetRef {
+        final Expression keyExpr;
+
+        RedisGetRef(Expression keyExpr) {
+            this.keyExpr = keyExpr;
+        }
+    }
+}

src/main/java/com/codechecker/cache/diff/ChangeType.java

@@ -7,6 +7,7 @@ public enum ChangeType {
     FIELD_REMOVED(Severity.P0, "字段删除"),
     TYPE_CHANGED(Severity.P0, "字段类型变更"),
     WRAPPER_ADDED(Severity.P0, "新增包装层"),
+    WRAPPER_REMOVED(Severity.P0, "删除包装层"),
     FIELD_PATH_MOVED(Severity.P0, "字段路径迁移"),
     FIELD_ADDED(Severity.P1, "新增字段"),
     KEY_PATTERN_CHANGED(Severity.P1, "Key 模式变更"),

src/main/java/com/codechecker/cache/diff/SchemaDiffer.java

@@ -22,40 +22,57 @@ public class SchemaDiffer {
     public List<SchemaChange> diff(TypeSchema oldSchema, TypeSchema newSchema) {
         List<SchemaChange> changes = new ArrayList<>();
 
-        Map<String, JsonType> oldLeaves = leaves(oldSchema);
+        Map<String, FieldSchema> oldAll = fieldsByPath(oldSchema);
-        Map<String, JsonType> newLeaves = leaves(newSchema);
+        Map<String, FieldSchema> newAll = fieldsByPath(newSchema);
+
+        Map<String, FieldSchema> oldLeaves = leaves(oldSchema);
+        Map<String, FieldSchema> newLeaves = leaves(newSchema);
+
+        Set<String> typeChangedPaths = new LinkedHashSet<>();
+
+        // 类型变更：同路径下 JsonType 或 Java 类型任一变化（覆盖 BigDecimal→String 等同桶变更）
+        for (String path : oldAll.keySet()) {
+            FieldSchema oldField = oldAll.get(path);
+            FieldSchema newField = newAll.get(path);
+            if (newField == null) {
+                continue;
+            }
+            if (!isTypeChange(oldField, newField)) {
+                continue;
+            }
+            typeChangedPaths.add(path);
+            String oldLabel = typeLabel(oldField);
+            String newLabel = typeLabel(newField);
+            SchemaChange c = new SchemaChange(ChangeType.TYPE_CHANGED);
+            c.setFieldPath(path);
+            c.setOldValue(oldLabel);
+            c.setNewValue(newLabel);
+            c.setMessage("字段 " + path + " 类型由 " + oldLabel + " 变为 " + newLabel);
+            changes.add(c);
+        }
 
         Set<String> removed = new LinkedHashSet<>(oldLeaves.keySet());
         removed.removeAll(newLeaves.keySet());
+        removed.removeAll(typeChangedPaths);
         Set<String> added = new LinkedHashSet<>(newLeaves.keySet());
         added.removeAll(oldLeaves.keySet());
+        added.removeAll(typeChangedPaths);
 
-        // 类型变更（同路径）
+        // 路径迁移：
-        for (String path : oldLeaves.keySet()) {
+        // - 下沉（加包装）：旧路径是新路径后缀，如 dbName → vo.dbName
-            if (newLeaves.containsKey(path)) {
+        // - 上提（拆包装）：新路径是旧路径后缀，如 vo.dbName → dbName
-                JsonType oldType = oldLeaves.get(path);
-                JsonType newType = newLeaves.get(path);
-                if (oldType != newType) {
-                    SchemaChange c = new SchemaChange(ChangeType.TYPE_CHANGED);
-                    c.setFieldPath(path);
-                    c.setOldValue(oldType.name());
-                    c.setNewValue(newType.name());
-                    c.setMessage("字段 " + path + " 类型由 " + oldType + " 变为 " + newType);
-                    changes.add(c);
-                }
-            }
-        }
-
-        // 路径迁移检测（removed 的路径是某 added 路径的后缀）
         List<String[]> moves = new ArrayList<>();
         Set<String> matchedRemoved = new LinkedHashSet<>();
         Set<String> matchedAdded = new LinkedHashSet<>();
         for (String r : removed) {
+            if (matchedRemoved.contains(r)) {
+                continue;
+            }
             for (String a : added) {
                 if (matchedAdded.contains(a)) {
                     continue;
                 }
-                if (isSuffix(a, r)) {
+                if (isSuffix(a, r) || isSuffix(r, a)) {
                     moves.add(new String[]{r, a});
                     matchedRemoved.add(r);
                     matchedAdded.add(a);
@@ -64,16 +81,16 @@ public class SchemaDiffer {
             }
         }
 
-        // 包装层检测：多个迁移共享同一新前缀
+        // 新增包装层：多个迁移共享同一新前缀
-        Map<String, Integer> prefixCount = new LinkedHashMap<>();
+        Map<String, Integer> newPrefixCount = new LinkedHashMap<>();
         for (String[] move : moves) {
             String a = move[1];
             if (a.contains(".")) {
                 String prefix = a.substring(0, a.indexOf('.'));
-                prefixCount.merge(prefix, 1, Integer::sum);
+                newPrefixCount.merge(prefix, 1, Integer::sum);
             }
         }
-        for (Map.Entry<String, Integer> e : prefixCount.entrySet()) {
+        for (Map.Entry<String, Integer> e : newPrefixCount.entrySet()) {
             if (e.getValue() >= 2) {
                 SchemaChange c = new SchemaChange(ChangeType.WRAPPER_ADDED);
                 c.setFieldPath(e.getKey());
@@ -83,6 +100,25 @@ public class SchemaDiffer {
             }
         }
 
+        // 删除包装层：多个迁移共享同一旧前缀
+        Map<String, Integer> oldPrefixCount = new LinkedHashMap<>();
+        for (String[] move : moves) {
+            String r = move[0];
+            if (r.contains(".")) {
+                String prefix = r.substring(0, r.indexOf('.'));
+                oldPrefixCount.merge(prefix, 1, Integer::sum);
+            }
+        }
+        for (Map.Entry<String, Integer> e : oldPrefixCount.entrySet()) {
+            if (e.getValue() >= 2) {
+                SchemaChange c = new SchemaChange(ChangeType.WRAPPER_REMOVED);
+                c.setFieldPath(e.getKey());
+                c.setOldValue(e.getKey());
+                c.setMessage("删除包装层 " + e.getKey() + "，其下字段被提升为外层（影响 " + e.getValue() + " 个字段）");
+                changes.add(c);
+            }
+        }
+
         for (String[] move : moves) {
             SchemaChange c = new SchemaChange(ChangeType.FIELD_PATH_MOVED);
             c.setFieldPath(move[1]);
@@ -99,7 +135,7 @@ public class SchemaDiffer {
             }
             SchemaChange c = new SchemaChange(ChangeType.FIELD_REMOVED);
             c.setFieldPath(r);
-            c.setOldValue(oldLeaves.get(r).name());
+            c.setOldValue(typeLabel(oldLeaves.get(r)));
             c.setMessage("删除字段 " + r);
             changes.add(c);
         }
@@ -111,7 +147,7 @@ public class SchemaDiffer {
             }
             SchemaChange c = new SchemaChange(ChangeType.FIELD_ADDED);
             c.setFieldPath(a);
-            c.setNewValue(newLeaves.get(a).name());
+            c.setNewValue(typeLabel(newLeaves.get(a)));
             c.setMessage("新增字段 " + a);
             changes.add(c);
         }
@@ -119,11 +155,104 @@ public class SchemaDiffer {
         return changes;
     }
 
-    private Map<String, JsonType> leaves(TypeSchema schema) {
+    private boolean isTypeChange(FieldSchema oldField, FieldSchema newField) {
-        Map<String, JsonType> result = new LinkedHashMap<>();
+        if (oldField.getJsonType() != newField.getJsonType()) {
+            return true;
+        }
+        String oldJava = normalizeJavaType(oldField.getJavaType());
+        String newJava = normalizeJavaType(newField.getJavaType());
+        if (oldJava.isEmpty() || newJava.isEmpty()) {
+            return false;
+        }
+        return !oldJava.equals(newJava);
+    }
+
+    /**
+     * 展示用类型标签：优先 Java 简单名，并附带 JsonType（便于识别同桶变更）。
+     */
+    private String typeLabel(FieldSchema field) {
+        if (field == null) {
+            return "?";
+        }
+        String java = simpleJavaType(field.getJavaType());
+        JsonType json = field.getJsonType();
+        if (java.isEmpty() || "?".equals(java)) {
+            return json == null ? "?" : json.name();
+        }
+        if (json == null) {
+            return java;
+        }
+        return java + "/" + json.name();
+    }
+
+    /**
+     * 归一化后用于相等比较：去掉泛型与包名，基本类型与包装类型视为同一序列化类型。
+     */
+    static String normalizeJavaType(String javaType) {
+        String simple = simpleJavaType(javaType);
+        if (simple.isEmpty() || "?".equals(simple)) {
+            return "";
+        }
+        switch (simple) {
+            case "int":
+                return "Integer";
+            case "long":
+                return "Long";
+            case "short":
+                return "Short";
+            case "byte":
+                return "Byte";
+            case "double":
+                return "Double";
+            case "float":
+                return "Float";
+            case "boolean":
+                return "Boolean";
+            case "char":
+                return "Character";
+            default:
+                return simple;
+        }
+    }
+
+    private static String simpleJavaType(String javaType) {
+        if (javaType == null) {
+            return "";
+        }
+        String s = javaType.trim();
+        if (s.isEmpty()) {
+            return "";
+        }
+        int lt = s.indexOf('<');
+        if (lt > 0) {
+            s = s.substring(0, lt).trim();
+        }
+        int dot = s.lastIndexOf('.');
+        if (dot >= 0 && dot < s.length() - 1) {
+            s = s.substring(dot + 1);
+        }
+        return s;
+    }
+
+    private Map<String, FieldSchema> fieldsByPath(TypeSchema schema) {
+        Map<String, FieldSchema> result = new LinkedHashMap<>();
+        if (schema == null) {
+            return result;
+        }
+        for (FieldSchema f : schema.getFields().values()) {
+            result.put(f.getPath(), f);
+        }
+        return result;
+    }
+
+    private Map<String, FieldSchema> leaves(TypeSchema schema) {
+        Map<String, FieldSchema> result = new LinkedHashMap<>();
+        if (schema == null) {
+            return result;
+        }
         for (FieldSchema f : schema.getFields().values()) {
             if (f.getJsonType() != JsonType.OBJECT && f.getJsonType() != JsonType.ARRAY) {
-                result.put(f.getPath(), f.getJsonType());
+                result.put(f.getPath(), f);
             }
         }
         return result;

src/main/java/com/codechecker/cache/key/RedisKeyResolver.java

@@ -1,10 +1,12 @@
 package com.codechecker.cache.key;
 
 import com.codechecker.cache.schema.SourceIndex;
+import com.github.javaparser.Position;
 import com.github.javaparser.ast.body.ClassOrInterfaceDeclaration;
 import com.github.javaparser.ast.body.FieldDeclaration;
 import com.github.javaparser.ast.body.MethodDeclaration;
 import com.github.javaparser.ast.body.VariableDeclarator;
+import com.github.javaparser.ast.expr.AssignExpr;
 import com.github.javaparser.ast.expr.BinaryExpr;
 import com.github.javaparser.ast.expr.Expression;
 import com.github.javaparser.ast.expr.FieldAccessExpr;
@@ -17,6 +19,8 @@ import java.util.Optional;
 
 /**
  * 尽力将 Redis key 表达式静态推断为一个「key 模式」，动态部分以 {@code *} 表示。
+ * <p>
+ * 对局部变量名：先查同名静态常量 / 同名方法 return，再回溯方法内使用点之前最近一次声明或赋值。
  */
 public class RedisKeyResolver {
 
@@ -52,12 +56,17 @@ public class RedisKeyResolver {
         }
         if (expr instanceof NameExpr) {
             String name = ((NameExpr) expr).getNameAsString();
+            // 1) 静态常量 2) 同名方法 return 3) 局部变量定值（不改变既有优先级）
             String constVal = lookupConstant(enclosingClass, name, context, depth);
             if (constVal != null) {
                 return constVal;
             }
             String methodVal = lookupMethodReturn(enclosingClass, name, context, depth);
-            return methodVal != null ? methodVal : "*";
+            if (methodVal != null) {
+                return methodVal;
+            }
+            String localVal = lookupLocalBinding(expr, name, enclosingClass, context, depth);
+            return localVal != null ? localVal : "*";
         }
         if (expr instanceof FieldAccessExpr) {
             FieldAccessExpr fae = (FieldAccessExpr) expr;
@@ -94,6 +103,76 @@ public class RedisKeyResolver {
         return "*";
     }
 
+    /**
+     * 在包围方法内，取使用点之前对 {@code name} 最近一次声明初始化或赋值的右侧表达式。
+     */
+    private String lookupLocalBinding(Expression useSite, String name,
+                                      ClassOrInterfaceDeclaration enclosingClass,
+                                      SourceIndex.IndexedType context, int depth) {
+        if (useSite == null || name == null || name.isEmpty()) {
+            return null;
+        }
+        MethodDeclaration method = useSite.findAncestor(MethodDeclaration.class).orElse(null);
+        if (method == null || !method.getBody().isPresent()) {
+            return null;
+        }
+
+        Position usePos = useSite.getBegin().orElse(null);
+        Expression bestRhs = null;
+        Position bestPos = null;
+
+        for (VariableDeclarator var : method.findAll(VariableDeclarator.class)) {
+            if (!name.equals(var.getNameAsString()) || !var.getInitializer().isPresent()) {
+                continue;
+            }
+            Position pos = var.getBegin().orElse(null);
+            if (!isUsableBinding(pos, usePos, bestPos)) {
+                continue;
+            }
+            bestPos = pos;
+            bestRhs = var.getInitializer().get();
+        }
+
+        for (AssignExpr assign : method.findAll(AssignExpr.class)) {
+            if (!(assign.getTarget() instanceof NameExpr)) {
+                continue;
+            }
+            if (!name.equals(((NameExpr) assign.getTarget()).getNameAsString())) {
+                continue;
+            }
+            Position pos = assign.getBegin().orElse(null);
+            if (!isUsableBinding(pos, usePos, bestPos)) {
+                continue;
+            }
+            bestPos = pos;
+            bestRhs = assign.getValue();
+        }
+
+        if (bestRhs == null) {
+            return null;
+        }
+        // 避免自引用（String key = key）陷入无意义递归
+        if (bestRhs instanceof NameExpr
+                && name.equals(((NameExpr) bestRhs).getNameAsString())) {
+            return null;
+        }
+        return resolveExpr(bestRhs, enclosingClass, context, depth + 1);
+    }
+
+    /** 候选须严格在使用点之前，并在多个候选中取最近一次。 */
+    private static boolean isUsableBinding(Position candidate, Position usePos, Position bestPos) {
+        if (candidate == null) {
+            return bestPos == null;
+        }
+        if (usePos != null && !candidate.isBefore(usePos)) {
+            return false;
+        }
+        if (bestPos == null) {
+            return true;
+        }
+        return candidate.isAfter(bestPos);
+    }
+
     private String lookupConstant(ClassOrInterfaceDeclaration clazz, String name,
                                   SourceIndex.IndexedType context, int depth) {
         if (clazz == null) {

src/main/java/com/codechecker/cache/report/ReportBuilder.java

@@ -1,5 +1,6 @@
 package com.codechecker.cache.report;
 
+import com.codechecker.cache.diff.ChangeType;
 import com.codechecker.cache.diff.SchemaChange;
 import com.codechecker.cache.diff.Severity;
 
@@ -14,7 +15,7 @@ import java.util.Set;
 /**
  * 将 {@link CheckReport} 渲染为企微 Markdown / 控制台文本。
  * <ul>
- *   <li>企微：按 key 展示位置/类型/序列化骨架变更，不含字段明细，不分 P0/P1/P2</li>
+ *   <li>企微：按 key 展示位置/类型/序列化骨架变更；类型变更另附摘要行，不分 P0/P1/P2</li>
  *   <li>未解析 key 展示源码表达式 + 灰色「key 无法解析」提示</li>
  *   <li>多 key 优先拼成一条；超过企微上限则按 key 拆成多条</li>
  *   <li>CI：先打字段明细，再完整输出企微 Markdown（拆分后的每条）</li>
@@ -28,7 +29,7 @@ public class ReportBuilder {
     private final String titlePrefix;
 
     public ReportBuilder(String titlePrefix) {
-        this.titlePrefix = titlePrefix == null ? "[缓存结构变更]" : titlePrefix;
+        this.titlePrefix = titlePrefix == null ? "[序列化结构变更]" : titlePrefix;
     }
 
     /**
@@ -63,7 +64,7 @@ public class ReportBuilder {
         String header = buildHeader(report);
         List<String> bodies = buildKeyBodies(report);
         if (bodies.isEmpty()) {
-            return Collections.singletonList(header + "未检测到缓存序列化结构变更。\n");
+            return Collections.singletonList(header + "未检测到序列化结构变更。\n");
         }
 
         StringBuilder combined = new StringBuilder(header);
@@ -128,25 +129,92 @@ public class ReportBuilder {
         String oldJson = nvl(kc.getOldSkeletonJson());
         String newJson = nvl(kc.getNewSkeletonJson());
         Set<String> oldHighlight = SkeletonAnnotator.pathsForOldSkeleton(kc.getFieldDetails());
-        Set<String> newHighlight = SkeletonAnnotator.pathsForNewSkeleton(kc.getFieldDetails());
+        Set<String> wrappersRemoved = SkeletonAnnotator.pathsForWrapperRemoved(kc.getFieldDetails());
+        Set<String> newGreen = SkeletonAnnotator.pathsForNewSkeleton(kc.getFieldDetails());
+        Set<String> wrappersAdded = SkeletonAnnotator.pathsForWrapperAdded(kc.getFieldDetails());
+        Set<String> typeGray = SkeletonAnnotator.pathsForTypeChanged(kc.getFieldDetails());
         String oldRendered = oldJson.isEmpty()
-                ? "" : "“" + SkeletonAnnotator.annotateOldForWecom(oldJson, oldHighlight) + "”";
+                ? "" : "“" + SkeletonAnnotator.annotateOldForWecom(
+                oldJson, oldHighlight, typeGray, wrappersRemoved) + "”";
         String newRendered = newJson.isEmpty()
-                ? "" : "“" + SkeletonAnnotator.annotateNewForWecom(newJson, newHighlight) + "”";
+                ? "" : "“" + SkeletonAnnotator.annotateNewForWecom(
+                newJson, newGreen, typeGray, wrappersAdded) + "”";
         if (oldJson.isEmpty() && !newJson.isEmpty()) {
-            sb.append("  > **value 新增为：** ").append(newRendered).append("\n\n");
+            sb.append("  > **value 新增为：** ").append(newRendered).append('\n');
         } else if (!oldJson.isEmpty() && newJson.isEmpty()) {
-            sb.append("  > **value 原结构：** ").append(oldRendered).append("（已删除写入）\n\n");
+            sb.append("  > **value 原结构：** ").append(oldRendered).append("（已删除写入）\n");
         } else {
             sb.append("  > **value值由：** ").append(oldRendered).append('\n');
-            sb.append("  > **变更为：** ").append(newRendered).append("\n\n");
+            sb.append("  > **变更为：** ").append(newRendered).append('\n');
         }
+        appendTypeChangeSummary(sb, kc.getFieldDetails());
+        sb.append('\n');
         return sb.toString();
     }
 
+    /**
+     * 类型变更单独一行摘要，避免仅靠骨架颜色/占位难以识别。
+     * 例：{@code > **类型变更**: amount <font color="warning">BigDecimal → Integer</font>}
+     */
+    private void appendTypeChangeSummary(StringBuilder sb, List<SchemaChange> details) {
+        if (details == null || details.isEmpty()) {
+            return;
+        }
+        List<String> parts = new ArrayList<>();
+        for (SchemaChange c : details) {
+            if (c == null || c.getChangeType() != ChangeType.TYPE_CHANGED) {
+                continue;
+            }
+            String path = nvl(c.getFieldPath());
+            if (path.isEmpty()) {
+                continue;
+            }
+            String oldType = displayJavaType(c.getOldValue());
+            String newType = displayJavaType(c.getNewValue());
+            StringBuilder part = new StringBuilder();
+            // 字段名用普通文本（避免反引号被企微渲染成色块）
+            part.append(path);
+            if (!oldType.isEmpty() || !newType.isEmpty()) {
+                part.append(" <font color=\"warning\">")
+                        .append(oldType.isEmpty() ? "?" : oldType)
+                        .append(" → ")
+                        .append(newType.isEmpty() ? "?" : newType)
+                        .append("</font>");
+            }
+            parts.add(part.toString());
+        }
+        if (parts.isEmpty()) {
+            return;
+        }
+        sb.append("  > **类型变更**: ");
+        for (int i = 0; i < parts.size(); i++) {
+            if (i > 0) {
+                sb.append("；");
+            }
+            sb.append(parts.get(i));
+        }
+        sb.append('\n');
+    }
+
+    /** 将 {@code BigDecimal/NUMBER} 转为展示用 {@code BigDecimal}。 */
+    private static String displayJavaType(String typeLabel) {
+        if (typeLabel == null) {
+            return "";
+        }
+        String s = typeLabel.trim();
+        if (s.isEmpty()) {
+            return "";
+        }
+        int slash = s.indexOf('/');
+        if (slash > 0) {
+            s = s.substring(0, slash).trim();
+        }
+        return s;
+    }
+
     /**
      * Key 行：展示解析后的模式或未解析时的源码表达式；未解析时追加灰色提示。
-     * 反引号仅包裹 key 文本，避免与加粗/颜色嵌套冲突。
+     * 反引号内仍须转义 {@code *}，否则企微会把 {@code *:*} 当成斜体吃掉通配符。
      */
     private void appendKeyLine(StringBuilder sb, String displayKey, String keyExpression,
                                boolean unresolved) {
@@ -156,17 +224,28 @@ public class ReportBuilder {
         if (keyText.isEmpty()) {
             keyText = "unknown-key";
         }
-        sb.append("- Key --> `").append(keyText).append('`');
+        sb.append("- Key --> `").append(escapeWeComCode(keyText)).append('`');
         if (unresolved) {
             sb.append(" <font color=\"comment\">（key 无法解析）</font>");
         }
         sb.append('\n');
     }
 
-    /** 位置、类型作为每个 Key 块的通用项。 */
+    /**
+     * 企微 markdown 代码片段内的转义：{@code *} 会触发斜体（即使包在反引号里）。
+     */
+    static String escapeWeComCode(String text) {
+        if (text == null || text.isEmpty()) {
+            return "";
+        }
+        // 反斜杠转义对企微不可靠；用全角 ＊ 保留通配语义且不被吃掉
+        return text.replace("*", "＊");
+    }
+
+    /** 位置、类型作为每个 Key 块的通用项（不加反引号，避免企微渲染成色块）。 */
     private void appendMetaLines(StringBuilder sb, String writeLocation, String valueType) {
-        sb.append("  > **位置**: `").append(nvl(writeLocation)).append("`\n");
+        sb.append("  > **位置**: ").append(nvl(writeLocation)).append('\n');
-        sb.append("  > **类型**: `").append(nvl(valueType)).append("`\n");
+        sb.append("  > **类型**: ").append(nvl(valueType)).append('\n');
     }
 
     private boolean isUnresolvedKey(String keyPattern) {
@@ -182,7 +261,7 @@ public class ReportBuilder {
     public String toConsole(CheckReport report) {
         StringBuilder sb = new StringBuilder();
         if (!report.hasChanges()) {
-            sb.append("未检测到缓存序列化结构变更。\n");
+            sb.append("未检测到序列化结构变更。\n");
             return sb.toString();
         }
 
@@ -202,7 +281,7 @@ public class ReportBuilder {
             for (SchemaChange c : list) {
                 sb.append("- **").append(c.getChangeType().getLabel()).append("**");
                 if (c.getKeyPattern() != null) {
-                    sb.append(" `").append(c.getKeyPattern()).append('`');
+                    sb.append(" `").append(escapeWeComCode(c.getKeyPattern())).append('`');
                 }
                 sb.append('\n');
                 if (c.getWriteLocation() != null) {
@@ -250,8 +329,8 @@ public class ReportBuilder {
                     + trimmed.substring(colonEn + 1).trim();
         }
         String[] knownPrefixes = {
-                "删除字段 ", "新增字段 ", "字段路径迁移 ", "新增包装层 ", "新增缓存写入点，",
+                "删除字段 ", "新增字段 ", "字段路径迁移 ", "新增包装层 ", "删除包装层 ",
-                "删除缓存写入点，原 value 类型: "
+                "新增缓存写入点，", "删除缓存写入点，原 value 类型: "
         };
         for (String prefix : knownPrefixes) {
             if (trimmed.startsWith(prefix)) {

src/main/java/com/codechecker/cache/report/SkeletonAnnotator.java

@@ -12,23 +12,28 @@ import java.util.Set;
 /**
  * 在骨架 JSON 中为改动字段加企微颜色标注（仅改动片段染色，其余明文）。
  * <ul>
- *   <li>删除字段 → 旧骨架，橙色 {@code warning}</li>
+ *   <li>删除包装层 → 旧骨架优先整段标橙（先于子路径）</li>
- *   <li>新增字段 / 新增包装层 → 新骨架，绿色 {@code info}</li>
+ *   <li>删除字段 / 路径迁移旧侧 → 旧骨架，橙色 {@code warning}</li>
- *   <li>路径迁移 → 旧路径橙、新路径绿</li>
+ *   <li>新增包装层 → 新骨架优先整体标绿（先于子路径）</li>
+ *   <li>新增字段 / 路径迁移新侧 → 新骨架，绿色 {@code info}</li>
+ *   <li>类型变更（修改类）→ 新旧骨架均为灰色 {@code comment}</li>
  * </ul>
  */
 final class SkeletonAnnotator {
 
-    /** 企微橙：删除 / 旧侧变更 */
+    /** 企微橙：删除 / 路径迁移旧侧 / 删除包装层 */
     static final String COLOR_REMOVE = "warning";
-    /** 企微绿：新增 / 新侧变更 */
+    /** 企微绿：新增 / 新侧路径迁移 */
     static final String COLOR_ADD = "info";
+    /** 企微灰：类型变更（修改类） */
+    static final String COLOR_TYPE = "comment";
 
     private static final String FONT_CLOSE = "</font>";
 
     private SkeletonAnnotator() {
     }
 
+    /** 旧骨架应橙色标注的路径：删除、路径迁移旧侧（不含类型变更与包装层删除）。 */
     static Set<String> pathsForOldSkeleton(List<SchemaChange> details) {
         Set<String> paths = new LinkedHashSet<>();
         if (details == null) {
@@ -47,6 +52,35 @@ final class SkeletonAnnotator {
         return paths;
     }
 
+    /** 删除包装层路径（旧骨架优先整段标橙）。 */
+    static Set<String> pathsForWrapperRemoved(List<SchemaChange> details) {
+        Set<String> paths = new LinkedHashSet<>();
+        if (details == null) {
+            return paths;
+        }
+        for (SchemaChange c : details) {
+            if (c != null && c.getChangeType() == ChangeType.WRAPPER_REMOVED) {
+                addIfPresent(paths, c.getFieldPath());
+            }
+        }
+        return paths;
+    }
+
+    /** 新增包装层路径（新骨架优先整段标绿）。 */
+    static Set<String> pathsForWrapperAdded(List<SchemaChange> details) {
+        Set<String> paths = new LinkedHashSet<>();
+        if (details == null) {
+            return paths;
+        }
+        for (SchemaChange c : details) {
+            if (c != null && c.getChangeType() == ChangeType.WRAPPER_ADDED) {
+                addIfPresent(paths, c.getFieldPath());
+            }
+        }
+        return paths;
+    }
+
+    /** 新骨架应绿色标注的路径：新增字段、路径迁移新侧（不含包装层与类型变更）。 */
     static Set<String> pathsForNewSkeleton(List<SchemaChange> details) {
         Set<String> paths = new LinkedHashSet<>();
         if (details == null) {
@@ -56,8 +90,7 @@ final class SkeletonAnnotator {
             if (c == null || c.getChangeType() == null) {
                 continue;
             }
-            if (c.getChangeType() == ChangeType.FIELD_ADDED
+            if (c.getChangeType() == ChangeType.FIELD_ADDED) {
-                    || c.getChangeType() == ChangeType.WRAPPER_ADDED) {
                 addIfPresent(paths, c.getFieldPath());
             } else if (c.getChangeType() == ChangeType.FIELD_PATH_MOVED) {
                 addIfPresent(paths, c.getFieldPath());
@@ -69,23 +102,68 @@ final class SkeletonAnnotator {
         return paths;
     }
 
-    /**
+    /** 类型变更路径：新旧骨架均用灰色标注。 */
-     * 标注旧骨架改动字段（删除 → 橙色）。
+    static Set<String> pathsForTypeChanged(List<SchemaChange> details) {
-     */
+        Set<String> paths = new LinkedHashSet<>();
-    static String annotateOldForWecom(String json, Set<String> highlightPaths) {
+        if (details == null) {
-        return annotateForWecom(json, highlightPaths, COLOR_REMOVE);
+            return paths;
+        }
+        for (SchemaChange c : details) {
+            if (c != null && c.getChangeType() == ChangeType.TYPE_CHANGED) {
+                addIfPresent(paths, c.getFieldPath());
+            }
+        }
+        return paths;
     }
 
     /**
-     * 标注新骨架改动字段（新增 → 绿色）。
+     * 标注旧骨架：删除等橙色；类型变更灰色。
      */
-    static String annotateNewForWecom(String json, Set<String> highlightPaths) {
+    static String annotateOldForWecom(String json, Set<String> orangePaths) {
-        return annotateForWecom(json, highlightPaths, COLOR_ADD);
+        return annotateOldForWecom(json, orangePaths, null, null);
+    }
+
+    static String annotateOldForWecom(String json, Set<String> orangePaths, Set<String> grayPaths) {
+        return annotateOldForWecom(json, orangePaths, grayPaths, null);
+    }
+
+    /**
+     * @param wrapperRemovedPaths 删除的包装层，优先于普通橙路径标注
+     */
+    static String annotateOldForWecom(String json, Set<String> orangePaths,
+                                      Set<String> grayPaths, Set<String> wrapperRemovedPaths) {
+        String result = annotateForWecom(json, grayPaths, COLOR_TYPE);
+        result = annotateForWecom(result, wrapperRemovedPaths, COLOR_REMOVE);
+        return annotateForWecom(result, orangePaths, COLOR_REMOVE);
+    }
+
+    /**
+     * 标注新骨架：类型变更灰色，新增绿色。
+     */
+    static String annotateNewForWecom(String json, Set<String> greenPaths) {
+        return annotateNewForWecom(json, greenPaths, null, null);
+    }
+
+    /**
+     * 标注新骨架（类型变更 → 灰色；包装层先绿；其余新增/迁移 → 绿）。
+     */
+    static String annotateNewForWecom(String json, Set<String> greenPaths, Set<String> grayTypePaths) {
+        return annotateNewForWecom(json, greenPaths, grayTypePaths, null);
+    }
+
+    /**
+     * @param wrapperPaths 包装层路径，优先于 {@code greenPaths} 标注，使 {@code "vo":{...}} 能整段染色
+     */
+    static String annotateNewForWecom(String json, Set<String> greenPaths,
+                                      Set<String> grayTypePaths, Set<String> wrapperPaths) {
+        String result = annotateForWecom(json, grayTypePaths, COLOR_TYPE);
+        result = annotateForWecom(result, wrapperPaths, COLOR_ADD);
+        return annotateForWecom(result, greenPaths, COLOR_ADD);
     }
 
     /**
      * 仅标注改动字段：其余正文保持普通文本。
-     * 改动片段格式：{@code <font color="warning|info">"field":value</font>}
+     * 改动片段格式：{@code <font color="warning|info|comment">"field":value</font>}
      */
     static String annotateForWecom(String json, Set<String> highlightPaths) {
         return annotateForWecom(json, highlightPaths, COLOR_REMOVE);

src/main/java/com/codechecker/cache/schema/JavaSchemaExtractor.java

@@ -21,12 +21,11 @@ public class JavaSchemaExtractor {
 
     private static final Set<String> STRING_TYPES = new HashSet<>(Arrays.asList(
             "String", "CharSequence", "char", "Character", "UUID",
-            "Date", "LocalDate", "LocalDateTime", "LocalTime", "Instant", "Timestamp",
+            "Date", "LocalDate", "LocalDateTime", "LocalTime", "Instant", "Timestamp"));
-            "BigDecimal"));
     private static final Set<String> NUMBER_TYPES = new HashSet<>(Arrays.asList(
             "int", "long", "short", "byte", "double", "float",
             "Integer", "Long", "Short", "Byte", "Double", "Float",
-            "Number", "BigInteger", "AtomicInteger", "AtomicLong"));
+            "Number", "BigInteger", "BigDecimal", "AtomicInteger", "AtomicLong"));
     private static final Set<String> BOOLEAN_TYPES = new HashSet<>(Arrays.asList(
             "boolean", "Boolean"));
     private static final Set<String> COLLECTION_TYPES = new HashSet<>(Arrays.asList(

src/main/java/com/codechecker/cache/schema/SkeletonJsonRenderer.java

@@ -45,7 +45,7 @@ public class SkeletonJsonRenderer {
     private Node buildTree(TypeSchema schema) {
         Node root = new Node(JsonType.OBJECT);
         for (FieldSchema field : schema.getFields().values()) {
-            putPath(root, field.getPath(), field.getJsonType());
+            putPath(root, field.getPath(), field.getJsonType(), field.getJavaType());
         }
         // 根数组：字段以 [] / [].xxx 记录
         if (root.children.size() == 1 && root.children.containsKey("[]")) {
@@ -56,7 +56,7 @@ public class SkeletonJsonRenderer {
         return root;
     }
 
-    private void putPath(Node root, String path, JsonType type) {
+    private void putPath(Node root, String path, JsonType type, String javaType) {
         List<Seg> segs = parsePath(path);
         if (segs.isEmpty()) {
             return;
@@ -70,10 +70,9 @@ public class SkeletonJsonRenderer {
                 arr.type = JsonType.ARRAY;
                 Node elem = arr.children.computeIfAbsent("[]", k -> new Node(JsonType.OBJECT));
                 if (last) {
-                    if (type == JsonType.OBJECT || type == JsonType.ARRAY || type == JsonType.MAP) {
+                    elem.type = type;
-                        elem.type = type;
+                    elem.javaType = javaType;
-                    } else {
+                    if (type != JsonType.OBJECT && type != JsonType.ARRAY && type != JsonType.MAP) {
-                        elem.type = type;
                         elem.leaf = true;
                     }
                 }
@@ -83,6 +82,7 @@ public class SkeletonJsonRenderer {
                         k -> new Node(last ? type : JsonType.OBJECT));
                 if (last) {
                     child.type = type;
+                    child.javaType = javaType;
                     child.leaf = type != JsonType.OBJECT && type != JsonType.ARRAY && type != JsonType.MAP;
                 } else if (child.type != JsonType.ARRAY) {
                     child.type = JsonType.OBJECT;
@@ -129,7 +129,7 @@ public class SkeletonJsonRenderer {
             return "[" + write(elem, elemPath, protectedPaths, compact) + "]";
         }
         if (node.leaf || isScalar(node.type)) {
-            return placeholder(node.type);
+            return placeholder(node.type, node.javaType);
         }
         if (node.type == JsonType.MAP) {
             return "{}";
@@ -161,7 +161,7 @@ public class SkeletonJsonRenderer {
             } else if (child.type == JsonType.ARRAY) {
                 sb.append(write(child, childPath, protectedPaths, compact));
             } else if (child.leaf || isScalar(child.type)) {
-                sb.append(placeholder(child.type));
+                sb.append(placeholder(child.type, child.javaType));
             } else {
                 sb.append(write(child, childPath, protectedPaths, compact));
             }
@@ -177,7 +177,7 @@ public class SkeletonJsonRenderer {
         if (child.type == JsonType.OBJECT || child.type == JsonType.MAP) {
             return "\"...\"";
         }
-        return placeholder(child.type);
+        return placeholder(child.type, child.javaType);
     }
 
     private boolean isProtectedUnder(String pathPrefix, Set<String> protectedPaths) {
@@ -327,13 +327,17 @@ public class SkeletonJsonRenderer {
                 || type == JsonType.BOOLEAN || type == JsonType.UNKNOWN;
     }
 
-    private String placeholder(JsonType type) {
+    /**
+     * 数字类型示意占位（合法 JSON number）：整数 {@code 0}、浮点 {@code 0.0}、BigDecimal {@code 0.00}。
+     * 仅用于通知可读性，不代表真实精度/scale。
+     */
+    private String placeholder(JsonType type, String javaType) {
         if (type == null) {
             return "null";
         }
         switch (type) {
             case NUMBER:
-                return "0";
+                return numberPlaceholder(javaType);
             case BOOLEAN:
                 return "false";
             case STRING:
@@ -349,6 +353,37 @@ public class SkeletonJsonRenderer {
         }
     }
 
+    private String numberPlaceholder(String javaType) {
+        String simple = simpleJavaType(javaType);
+        if ("BigDecimal".equals(simple)) {
+            return "0.00";
+        }
+        if ("float".equals(simple) || "Float".equals(simple)
+                || "double".equals(simple) || "Double".equals(simple)) {
+            return "0.0";
+        }
+        return "0";
+    }
+
+    private static String simpleJavaType(String javaType) {
+        if (javaType == null) {
+            return "";
+        }
+        String s = javaType.trim();
+        if (s.isEmpty()) {
+            return "";
+        }
+        int lt = s.indexOf('<');
+        if (lt > 0) {
+            s = s.substring(0, lt).trim();
+        }
+        int dot = s.lastIndexOf('.');
+        if (dot >= 0 && dot < s.length() - 1) {
+            s = s.substring(dot + 1);
+        }
+        return s;
+    }
+
     private String escape(String s) {
         return s.replace("\\", "\\\\").replace("\"", "\\\"");
     }
@@ -365,6 +400,7 @@ public class SkeletonJsonRenderer {
 
     private static final class Node {
         JsonType type;
+        String javaType;
         boolean leaf;
         final Map<String, Node> children = new LinkedHashMap<>();
 

src/main/resources/default-config.yaml

@@ -1,4 +1,4 @@
-# cache-schema-checker 内置默认配置
+# serialization-schema-checker 内置默认配置
 # 业务仓库通过 --config 指定的配置会与本文件深度合并（业务配置优先）。
 
 # 总开关：false 时不执行检测、不发通知、流水线直接通过
@@ -19,7 +19,7 @@ notify:
   enabled: true
   webhook_url: ""
   notify_on_clean: false
-  title_prefix: "[缓存结构变更]"
+  title_prefix: "[序列化结构变更]"
 
 # 忽略规则
 ignore:
@@ -44,7 +44,9 @@ detection:
     - W03   # stringRedisTemplate.opsForValue().set(key, JsonUtil.getObjectToString(x), ...)
     - W04   # redisTemplate.opsForValue().set(key, obj, ...)
     - W05   # redisTemplate.opsForHash().put(key, field, obj)
-  # 类型推断最低置信度，低于此值降级为 P2 提示
+  # W06 读侧辅助：用 parseObject / getJsonToBean 等补强写入点 value 类型（非写入模式）
+  read_hints_enabled: true
+  # 类型推断最低置信度，低于此值标记为低置信度提示
   min_confidence: 0.6
   # 字段展开最大深度（防止循环引用）
   max_field_depth: 8

src/test/java/com/codechecker/cache/detector/CacheReadHintDetectorTest.java

@@ -0,0 +1,71 @@
+package com.codechecker.cache.detector;
+
+import com.codechecker.cache.schema.SourceIndex;
+import org.junit.jupiter.api.Test;
+
+import java.util.List;
+
+import static org.junit.jupiter.api.Assertions.assertEquals;
+import static org.junit.jupiter.api.Assertions.assertFalse;
+import static org.junit.jupiter.api.Assertions.assertTrue;
+
+class CacheReadHintDetectorTest {
+
+    @Test
+    void detectsParseObjectWithRedisGetPairing() {
+        String vo = ""
+                + "package demo.model;\n"
+                + "public class FooVo { private String name; }\n";
+        String source = ""
+                + "package demo;\n"
+                + "import demo.model.FooVo;\n"
+                + "public class CacheService {\n"
+                + "  private RedisUtil redisUtil;\n"
+                + "  private static final String KEY = \"demo:foo:\";\n"
+                + "  public FooVo load(String id) {\n"
+                + "    String raw = redisUtil.getString(KEY + id);\n"
+                + "    return JSON.parseObject(raw, FooVo.class);\n"
+                + "  }\n"
+                + "  public void save(String id, FooVo vo) {\n"
+                + "    redisUtil.insert(KEY + id, JSON.toJSONString(vo), 60);\n"
+                + "  }\n"
+                + "}\n";
+
+        SourceIndex index = new SourceIndex();
+        index.addSource(vo);
+        index.addSource(source);
+
+        List<CacheReadHint> hints = new CacheReadHintDetector(index)
+                .detect("CacheService.java", source);
+        assertEquals(1, hints.size());
+        CacheReadHint hint = hints.get(0);
+        assertEquals("demo.model.FooVo", hint.getResolvedValueType());
+        assertFalse(hint.isRootArray());
+        assertEquals("demo:foo:*", hint.getResolvedKeyPattern());
+    }
+
+    @Test
+    void detectsParseArray() {
+        String vo = ""
+                + "package demo.model;\n"
+                + "public class ItemVo { private int rank; }\n";
+        String source = ""
+                + "package demo;\n"
+                + "import demo.model.ItemVo;\n"
+                + "public class ListCache {\n"
+                + "  public List<ItemVo> load(String raw) {\n"
+                + "    return JSON.parseArray(raw, ItemVo.class);\n"
+                + "  }\n"
+                + "}\n";
+
+        SourceIndex index = new SourceIndex();
+        index.addSource(vo);
+        index.addSource(source);
+
+        List<CacheReadHint> hints = new CacheReadHintDetector(index)
+                .detect("ListCache.java", source);
+        assertEquals(1, hints.size());
+        assertTrue(hints.get(0).isRootArray());
+        assertEquals("demo.model.ItemVo", hints.get(0).getResolvedValueType());
+    }
+}

src/test/java/com/codechecker/cache/diff/SchemaDifferTest.java

@@ -40,6 +40,41 @@ class SchemaDifferTest {
         assertEquals(2, moved, "dbName 与 linkList[].id 均应迁移");
     }
 
+    @Test
+    void detectsWrapperRemovalAndUpwardMoves() {
+        TypeSchema oldS = new TypeSchema("CacheEnvelope");
+        oldS.add(new FieldSchema("vo", JsonType.OBJECT, "TenantVO"));
+        oldS.add(new FieldSchema("vo.dbName", JsonType.STRING, "String"));
+        oldS.add(new FieldSchema("vo.linkList", JsonType.ARRAY, "List"));
+        oldS.add(new FieldSchema("vo.linkList[]", JsonType.OBJECT, "TenantLinkModel"));
+        oldS.add(new FieldSchema("vo.linkList[].id", JsonType.STRING, "String"));
+
+        TypeSchema newS = new TypeSchema("TenantVO");
+        newS.add(new FieldSchema("dbName", JsonType.STRING, "String"));
+        newS.add(new FieldSchema("linkList", JsonType.ARRAY, "List"));
+        newS.add(new FieldSchema("linkList[]", JsonType.OBJECT, "TenantLinkModel"));
+        newS.add(new FieldSchema("linkList[].id", JsonType.STRING, "String"));
+
+        List<SchemaChange> changes = new SchemaDiffer().diff(oldS, newS);
+        List<ChangeType> types = changes.stream().map(SchemaChange::getChangeType).collect(Collectors.toList());
+
+        assertTrue(types.contains(ChangeType.WRAPPER_REMOVED), "应检测到包装层删除");
+        assertTrue(types.contains(ChangeType.FIELD_PATH_MOVED), "应检测到字段上提迁移");
+        assertTrue(types.stream().noneMatch(t -> t == ChangeType.FIELD_ADDED),
+                "上提字段不应再被当成新增");
+        assertTrue(types.stream().noneMatch(t -> t == ChangeType.FIELD_REMOVED),
+                "上提字段不应再被当成删除");
+
+        SchemaChange wrapper = changes.stream()
+                .filter(c -> c.getChangeType() == ChangeType.WRAPPER_REMOVED)
+                .findFirst()
+                .orElseThrow(IllegalStateException::new);
+        assertEquals("vo", wrapper.getFieldPath());
+
+        long moved = changes.stream().filter(c -> c.getChangeType() == ChangeType.FIELD_PATH_MOVED).count();
+        assertEquals(2, moved, "vo.dbName / vo.linkList[].id 均应上提");
+    }
+
     @Test
     void detectsTypeChange() {
         TypeSchema oldS = new TypeSchema("A");
@@ -51,6 +86,57 @@ class SchemaDifferTest {
         assertEquals(1, changes.size());
         assertEquals(ChangeType.TYPE_CHANGED, changes.get(0).getChangeType());
         assertEquals(Severity.P0, changes.get(0).getSeverity());
+        assertTrue(changes.get(0).getMessage().contains("String"));
+        assertTrue(changes.get(0).getMessage().contains("Integer"));
+    }
+
+    @Test
+    void detectsBigDecimalToStringEvenWhenBothWereStringBucketHistorically() {
+        // 即便 JsonType 同为 STRING，javaType 变化也必须检出（用户真实漏报场景）
+        TypeSchema oldS = new TypeSchema("A");
+        oldS.add(new FieldSchema("amount", JsonType.STRING, "BigDecimal"));
+        TypeSchema newS = new TypeSchema("A");
+        newS.add(new FieldSchema("amount", JsonType.STRING, "String"));
+
+        List<SchemaChange> changes = new SchemaDiffer().diff(oldS, newS);
+        assertEquals(1, changes.size());
+        assertEquals(ChangeType.TYPE_CHANGED, changes.get(0).getChangeType());
+        assertEquals("amount", changes.get(0).getFieldPath());
+        assertTrue(changes.get(0).getOldValue().contains("BigDecimal"));
+        assertTrue(changes.get(0).getNewValue().contains("String"));
+    }
+
+    @Test
+    void detectsBigDecimalToStringViaJsonTypeAndJavaType() {
+        TypeSchema oldS = new TypeSchema("A");
+        oldS.add(new FieldSchema("amount", JsonType.NUMBER, "BigDecimal"));
+        TypeSchema newS = new TypeSchema("A");
+        newS.add(new FieldSchema("amount", JsonType.STRING, "String"));
+
+        List<SchemaChange> changes = new SchemaDiffer().diff(oldS, newS);
+        assertEquals(1, changes.size());
+        assertEquals(ChangeType.TYPE_CHANGED, changes.get(0).getChangeType());
+    }
+
+    @Test
+    void detectsIntegerToLongWithinNumberBucket() {
+        TypeSchema oldS = new TypeSchema("A");
+        oldS.add(new FieldSchema("id", JsonType.NUMBER, "Integer"));
+        TypeSchema newS = new TypeSchema("A");
+        newS.add(new FieldSchema("id", JsonType.NUMBER, "Long"));
+
+        List<SchemaChange> changes = new SchemaDiffer().diff(oldS, newS);
+        assertEquals(1, changes.size());
+        assertEquals(ChangeType.TYPE_CHANGED, changes.get(0).getChangeType());
+    }
+
+    @Test
+    void noChangeWhenPrimitiveAndWrapperEquivalent() {
+        TypeSchema a = new TypeSchema("A");
+        a.add(new FieldSchema("n", JsonType.NUMBER, "int"));
+        TypeSchema b = new TypeSchema("A");
+        b.add(new FieldSchema("n", JsonType.NUMBER, "Integer"));
+        assertTrue(new SchemaDiffer().diff(a, b).isEmpty());
     }
 
     @Test

src/test/java/com/codechecker/cache/key/RedisKeyResolverTest.java

@@ -65,4 +65,91 @@ class RedisKeyResolverTest {
                 buildCall, clazz, index.get("jnpf.util.TenantDbContentCacheHelper"));
         assertEquals("tenant:db:content:*", pattern);
     }
+
+    @Test
+    void resolvesLocalVariableAssignedFromStringFormat() {
+        String source = ""
+                + "package demo;\n"
+                + "public class CombinationRepo {\n"
+                + "  private static final String COMBINATION_QUERY_KEY = "
+                + "\"data_analysis:combination_query_key:%s:%s\";\n"
+                + "  public void write(String tenantId, String hashKey) {\n"
+                + "    String key = String.format(COMBINATION_QUERY_KEY, tenantId, hashKey);\n"
+                + "    redisSet(key);\n"
+                + "  }\n"
+                + "  void redisSet(String k) {}\n"
+                + "}\n";
+
+        SourceIndex index = new SourceIndex();
+        index.addSource(source);
+        CompilationUnit cu = StaticJavaParser.parse(source);
+        ClassOrInterfaceDeclaration clazz = cu.getType(0).asClassOrInterfaceDeclaration();
+        // redisSet(key) 的实参 key —— 局部变量使用点
+        MethodCallExpr redisSet = cu.findAll(MethodCallExpr.class).stream()
+                .filter(m -> "redisSet".equals(m.getNameAsString()))
+                .findFirst()
+                .orElseThrow(IllegalStateException::new);
+        Expression keyArg = redisSet.getArgument(0);
+
+        String pattern = new RedisKeyResolver(index).resolve(
+                keyArg, clazz, index.get("demo.CombinationRepo"));
+        assertEquals("data_analysis:combination_query_key:*:*", pattern);
+    }
+
+    @Test
+    void localReassignmentUsesNearestBindingBeforeUse() {
+        String source = ""
+                + "package demo;\n"
+                + "public class ReassignRepo {\n"
+                + "  private static final String A = \"prefix:a:%s\";\n"
+                + "  private static final String B = \"prefix:b:%s\";\n"
+                + "  public void write(String id) {\n"
+                + "    String key = String.format(A, id);\n"
+                + "    key = String.format(B, id);\n"
+                + "    redisSet(key);\n"
+                + "  }\n"
+                + "  void redisSet(String k) {}\n"
+                + "}\n";
+
+        SourceIndex index = new SourceIndex();
+        index.addSource(source);
+        CompilationUnit cu = StaticJavaParser.parse(source);
+        ClassOrInterfaceDeclaration clazz = cu.getType(0).asClassOrInterfaceDeclaration();
+        MethodCallExpr redisSet = cu.findAll(MethodCallExpr.class).stream()
+                .filter(m -> "redisSet".equals(m.getNameAsString()))
+                .findFirst()
+                .orElseThrow(IllegalStateException::new);
+
+        String pattern = new RedisKeyResolver(index).resolve(
+                redisSet.getArgument(0), clazz, index.get("demo.ReassignRepo"));
+        assertEquals("prefix:b:*", pattern);
+    }
+
+    @Test
+    void staticConstantStillPreferredOverLocalNameCollision() {
+        // 局部变量名与静态常量同名时：仍优先静态常量（保持旧语义）
+        String source = ""
+                + "package demo;\n"
+                + "public class Collision {\n"
+                + "  private static final String KEY = \"const:prefix:%s\";\n"
+                + "  public void write(String id) {\n"
+                + "    String KEY = \"local:\" + id;\n"
+                + "    redisSet(KEY);\n"
+                + "  }\n"
+                + "  void redisSet(String k) {}\n"
+                + "}\n";
+
+        SourceIndex index = new SourceIndex();
+        index.addSource(source);
+        CompilationUnit cu = StaticJavaParser.parse(source);
+        ClassOrInterfaceDeclaration clazz = cu.getType(0).asClassOrInterfaceDeclaration();
+        MethodCallExpr redisSet = cu.findAll(MethodCallExpr.class).stream()
+                .filter(m -> "redisSet".equals(m.getNameAsString()))
+                .findFirst()
+                .orElseThrow(IllegalStateException::new);
+
+        String pattern = new RedisKeyResolver(index).resolve(
+                redisSet.getArgument(0), clazz, index.get("demo.Collision"));
+        assertEquals("const:prefix:%s", pattern);
+    }
 }

src/test/java/com/codechecker/cache/report/ReportBuilderTest.java

@@ -5,6 +5,7 @@ import com.codechecker.cache.diff.SchemaChange;
 import com.codechecker.cache.diff.Severity;
 import org.junit.jupiter.api.Test;
 
+import java.util.ArrayList;
 import java.util.Collections;
 import java.util.LinkedHashSet;
 import java.util.List;
@@ -56,12 +57,12 @@ class ReportBuilderTest {
         key.getFieldDetails().add(addedErr);
         report.getKeyChanges().add(key);
 
-        String md = new ReportBuilder("[缓存结构变更]").toMarkdown(report);
+        String md = new ReportBuilder("[序列化结构变更]").toMarkdown(report);
 
         assertTrue(md.contains("- Key --> `saas:period-config:migration:current`"));
         assertFalse(md.contains("（key 无法解析）"));
-        assertTrue(md.contains("> **位置**: `SaasPeriodConfigMigrationRedisSupport#putCurrent:41`"));
+        assertTrue(md.contains("> **位置**: SaasPeriodConfigMigrationRedisSupport#putCurrent:41"));
-        assertTrue(md.contains("> **类型**: `MigrationCurrentVo`"));
+        assertTrue(md.contains("> **类型**: MigrationCurrentVo"));
 
         int oldSection = md.indexOf("> **value值由：**");
         int newSection = md.indexOf("> **变更为：**");
@@ -111,6 +112,174 @@ class ReportBuilderTest {
         assertFalse(newMd.startsWith("`"));
     }
 
+    @Test
+    void typeChangedHighlightedGrayInBothSkeletons() {
+        SchemaChange typeChanged = new SchemaChange(ChangeType.TYPE_CHANGED);
+        typeChanged.setFieldPath("amount");
+        typeChanged.setOldValue("BigDecimal/NUMBER");
+        typeChanged.setNewValue("String/STRING");
+        typeChanged.setMessage("字段 amount 类型由 BigDecimal/NUMBER 变为 String/STRING");
+
+        String oldJson = "{\"amount\":0,\"name\":\"\"}";
+        String newJson = "{\"amount\":\"\",\"name\":\"\"}";
+        List<SchemaChange> details = Collections.singletonList(typeChanged);
+
+        Set<String> oldPaths = SkeletonAnnotator.pathsForOldSkeleton(details);
+        Set<String> newGreen = SkeletonAnnotator.pathsForNewSkeleton(details);
+        Set<String> typeGray = SkeletonAnnotator.pathsForTypeChanged(details);
+
+        assertTrue(oldPaths.isEmpty());
+        assertTrue(newGreen.isEmpty());
+        assertEquals(Collections.singleton("amount"), typeGray);
+
+        String oldMd = SkeletonAnnotator.annotateOldForWecom(oldJson, oldPaths, typeGray);
+        String newMd = SkeletonAnnotator.annotateNewForWecom(newJson, newGreen, typeGray);
+
+        assertTrue(oldMd.contains("<font color=\"comment\">\"amount\":0</font>"));
+        assertFalse(oldMd.contains("<font color=\"warning\">\"amount\""));
+        assertFalse(oldMd.contains("<font color=\"comment\">\"name\":\"\"</font>"));
+        assertTrue(newMd.contains("<font color=\"comment\">\"amount\":\"\"</font>"));
+        assertFalse(newMd.contains("<font color=\"info\">\"amount\""));
+        assertFalse(newMd.contains("<font color=\"warning\">\"amount\""));
+        assertFalse(newMd.contains("<font color=\"comment\">\"name\":\"\"</font>"));
+
+        CheckReport report = baseReport();
+        KeyStructureChange key = new KeyStructureChange();
+        key.setKeyPattern("cache:amount");
+        key.setWriteLocation("Demo#put:1");
+        key.setValueType("AmountVo");
+        key.setOldSkeletonJson(oldJson);
+        key.setNewSkeletonJson(newJson);
+        key.getFieldDetails().add(typeChanged);
+        report.getKeyChanges().add(key);
+
+        String md = new ReportBuilder("[序列化结构变更]").toMarkdown(report);
+        int oldSection = md.indexOf("> **value值由：**");
+        int newSection = md.indexOf("> **变更为：**");
+        assertTrue(oldSection >= 0);
+        assertTrue(newSection > oldSection);
+        String oldPart = md.substring(oldSection, newSection);
+        String newPart = md.substring(newSection);
+        assertTrue(oldPart.contains("<font color=\"comment\">\"amount\":0</font>"));
+        assertTrue(newPart.contains("<font color=\"comment\">\"amount\":\"\"</font>"));
+        assertFalse(newPart.contains("<font color=\"info\">\"amount\""));
+        assertFalse(oldPart.contains("<font color=\"warning\">\"amount\""));
+        assertTrue(md.contains("> **类型变更**: amount <font color=\"warning\">BigDecimal → String</font>"),
+                "应包含类型变更摘要行，实际 markdown:\n" + md);
+    }
+
+    @Test
+    void wrapperAddedHighlightedBeforeNestedMovedFields() {
+        SchemaChange wrapper = new SchemaChange(ChangeType.WRAPPER_ADDED);
+        wrapper.setFieldPath("vo");
+        wrapper.setMessage("新增包装层 vo");
+
+        SchemaChange movedDb = new SchemaChange(ChangeType.FIELD_PATH_MOVED);
+        movedDb.setFieldPath("vo.dbName");
+        movedDb.setOldValue("dbName");
+        movedDb.setNewValue("vo.dbName");
+
+        SchemaChange movedLink = new SchemaChange(ChangeType.FIELD_PATH_MOVED);
+        movedLink.setFieldPath("vo.linkList[].id");
+        movedLink.setOldValue("linkList[].id");
+        movedLink.setNewValue("vo.linkList[].id");
+
+        List<SchemaChange> details = new ArrayList<>();
+        details.add(wrapper);
+        details.add(movedDb);
+        details.add(movedLink);
+
+        String oldJson = "{\"dbName\":\"\",\"linkList\":[{\"id\":\"\"}]}";
+        String newJson = "{\"vo\":{\"dbName\":\"\",\"linkList\":[{\"id\":\"\"}]}}";
+
+        Set<String> wrappers = SkeletonAnnotator.pathsForWrapperAdded(details);
+        Set<String> newGreen = SkeletonAnnotator.pathsForNewSkeleton(details);
+        assertEquals(Collections.singleton("vo"), wrappers);
+        assertTrue(newGreen.contains("vo.dbName"));
+        assertFalse(newGreen.contains("vo"), "包装层不应再混入普通新增路径");
+
+        // 先标包装层：整段 vo 对象应绿；子路径随后因已在 font 内可跳过
+        String newMd = SkeletonAnnotator.annotateNewForWecom(newJson, newGreen, null, wrappers);
+        assertTrue(newMd.contains("<font color=\"info\">\"vo\":{"), newMd);
+        assertTrue(newMd.contains("</font>"), newMd);
+
+        CheckReport report = baseReport();
+        KeyStructureChange key = new KeyStructureChange();
+        key.setKeyPattern("tenant:db:content:*");
+        key.setWriteLocation("TenantDbContentCacheHelper#cacheNotFound:62");
+        key.setValueType("TenantVO");
+        key.setOldSkeletonJson(oldJson);
+        key.setNewSkeletonJson(newJson);
+        key.getFieldDetails().addAll(details);
+        report.getKeyChanges().add(key);
+
+        String md = new ReportBuilder("[序列化结构变更]").toMarkdown(report);
+        int newSection = md.indexOf("> **变更为：**");
+        assertTrue(newSection >= 0);
+        String newPart = md.substring(newSection);
+        assertTrue(newPart.contains("<font color=\"info\">\"vo\":{"),
+                "包装层 vo 应整段绿色，实际:\n" + newPart);
+    }
+
+    @Test
+    void wrapperRemovedHighlightedBeforeNestedMovedFieldsOnOldSkeleton() {
+        SchemaChange wrapper = new SchemaChange(ChangeType.WRAPPER_REMOVED);
+        wrapper.setFieldPath("vo");
+        wrapper.setOldValue("vo");
+        wrapper.setMessage("删除包装层 vo");
+
+        SchemaChange movedDb = new SchemaChange(ChangeType.FIELD_PATH_MOVED);
+        movedDb.setFieldPath("dbName");
+        movedDb.setOldValue("vo.dbName");
+        movedDb.setNewValue("dbName");
+
+        SchemaChange movedLink = new SchemaChange(ChangeType.FIELD_PATH_MOVED);
+        movedLink.setFieldPath("linkList[].id");
+        movedLink.setOldValue("vo.linkList[].id");
+        movedLink.setNewValue("linkList[].id");
+
+        List<SchemaChange> details = new ArrayList<>();
+        details.add(wrapper);
+        details.add(movedDb);
+        details.add(movedLink);
+
+        String oldJson = "{\"vo\":{\"dbName\":\"\",\"linkList\":[{\"id\":\"\"}]}}";
+        String newJson = "{\"dbName\":\"\",\"linkList\":[{\"id\":\"\"}]}";
+
+        Set<String> wrappersRemoved = SkeletonAnnotator.pathsForWrapperRemoved(details);
+        Set<String> oldOrange = SkeletonAnnotator.pathsForOldSkeleton(details);
+        Set<String> newGreen = SkeletonAnnotator.pathsForNewSkeleton(details);
+
+        assertEquals(Collections.singleton("vo"), wrappersRemoved);
+        assertTrue(oldOrange.contains("vo.dbName"));
+        assertTrue(newGreen.contains("dbName"));
+        assertFalse(newGreen.contains("vo"));
+
+        String oldMd = SkeletonAnnotator.annotateOldForWecom(oldJson, oldOrange, null, wrappersRemoved);
+        assertTrue(oldMd.contains("<font color=\"warning\">\"vo\":{"), oldMd);
+
+        String newMd = SkeletonAnnotator.annotateNewForWecom(newJson, newGreen, null, null);
+        assertTrue(newMd.contains("<font color=\"info\">\"dbName\":\"\"</font>"), newMd);
+        assertFalse(newMd.contains("<font color=\"info\">\"vo\""), newMd);
+
+        CheckReport report = baseReport();
+        KeyStructureChange key = new KeyStructureChange();
+        key.setKeyPattern("tenant:db:content:＊");
+        key.setWriteLocation("TenantDbContentCacheHelper#cacheNotFound:62");
+        key.setValueType("TenantVO");
+        key.setOldSkeletonJson(oldJson);
+        key.setNewSkeletonJson(newJson);
+        key.getFieldDetails().addAll(details);
+        report.getKeyChanges().add(key);
+
+        String md = new ReportBuilder("[序列化结构变更]").toMarkdown(report);
+        int oldSection = md.indexOf("> **value值由：**");
+        int newSection = md.indexOf("> **变更为：**");
+        assertTrue(oldSection >= 0 && newSection > oldSection);
+        assertTrue(md.substring(oldSection, newSection).contains("<font color=\"warning\">\"vo\":{"),
+                "旧骨架应整段橙标 vo，实际:\n" + md);
+    }
+
     @Test
     void consoleContainsFieldDetailsAndWecomMarkdown() {
         CheckReport report = new CheckReport();
@@ -133,13 +302,13 @@ class ReportBuilderTest {
         key.getFieldDetails().add(detail);
         report.getKeyChanges().add(key);
 
-        String console = new ReportBuilder("[缓存结构变更]").toConsole(report);
+        String console = new ReportBuilder("[序列化结构变更]").toConsole(report);
         assertTrue(console.contains("======== 字段明细 ========"));
         assertTrue(console.contains("**删除字段**: x"));
         assertTrue(console.contains("<font color=\"warning\">\"x\":\"\"</font>"));
         assertFalse(console.contains("<font color=\"info\">\"x\":\"\"</font>"));
-        assertTrue(console.contains("> **位置**: `"));
+        assertTrue(console.contains("> **位置**: "));
-        assertTrue(console.contains("> **类型**: `"));
+        assertTrue(console.contains("> **类型**: "));
         assertTrue(console.contains("> **value值由：**"));
         assertTrue(console.contains("> **变更为：**"));
         assertFalse(console.contains("`{\"x\":\"\"}`"));
@@ -161,10 +330,10 @@ class ReportBuilderTest {
         key.getFieldDetails().add(added);
         report.getKeyChanges().add(key);
 
-        String md = new ReportBuilder("[缓存结构变更]").toMarkdown(report);
+        String md = new ReportBuilder("[序列化结构变更]").toMarkdown(report);
         assertTrue(md.contains("- Key --> `req.getKey()` <font color=\"comment\">（key 无法解析）</font>"));
-        assertTrue(md.contains("> **位置**: `ClockInXxxService#export:128`"));
+        assertTrue(md.contains("> **位置**: ClockInXxxService#export:128"));
-        assertTrue(md.contains("> **类型**: `List<ClockInExportVo>`"));
+        assertTrue(md.contains("> **类型**: List<ClockInExportVo>"));
         assertFalse(md.contains("`unknown-key`"));
     }
 
@@ -185,7 +354,7 @@ class ReportBuilderTest {
         report.getKeyChanges().add(simpleKey("k-a", "{\"a\":\"\"}", "{\"a\":\"\",\"x\":\"\"}", "x"));
         report.getKeyChanges().add(simpleKey("k-b", "{\"b\":\"\"}", "{\"b\":\"\",\"y\":\"\"}", "y"));
 
-        List<String> messages = new ReportBuilder("[缓存结构变更]").toWeComMessages(report);
+        List<String> messages = new ReportBuilder("[序列化结构变更]").toWeComMessages(report);
         assertEquals(1, messages.size());
         assertTrue(messages.get(0).contains("k-a"));
         assertTrue(messages.get(0).contains("k-b"));
@@ -200,7 +369,7 @@ class ReportBuilderTest {
         report.getKeyChanges().add(simpleKey("fat-key-1", fat, fat + "1", null));
         report.getKeyChanges().add(simpleKey("fat-key-2", fat, fat + "2", null));
 
-        List<String> messages = new ReportBuilder("[缓存结构变更]").toWeComMessages(report);
+        List<String> messages = new ReportBuilder("[序列化结构变更]").toWeComMessages(report);
         assertEquals(2, messages.size(), "超长应按 key 拆成 2 条");
         assertTrue(messages.get(0).contains("fat-key-1"));
         assertFalse(messages.get(0).contains("fat-key-2"));
@@ -209,13 +378,31 @@ class ReportBuilderTest {
         assertTrue(ReportBuilder.utf8Bytes(messages.get(0)) <= ReportBuilder.WECOM_MARKDOWN_MAX_BYTES);
         assertTrue(ReportBuilder.utf8Bytes(messages.get(1)) <= ReportBuilder.WECOM_MARKDOWN_MAX_BYTES);
         // 抬头在每条中重复
-        assertTrue(messages.get(0).contains("## [缓存结构变更] jnpf-java-cloud"));
+        assertTrue(messages.get(0).contains("## [序列化结构变更] jnpf-java-cloud"));
-        assertTrue(messages.get(1).contains("## [缓存结构变更] jnpf-java-cloud"));
+        assertTrue(messages.get(1).contains("## [序列化结构变更] jnpf-java-cloud"));
 
-        String console = new ReportBuilder("[缓存结构变更]").toConsole(report);
+        String console = new ReportBuilder("[序列化结构变更]").toConsole(report);
         assertTrue(console.contains("超长已按 key 拆为 2 条"));
     }
 
+    @Test
+    void wildcardAsteriskInKeyEscapedForWeCom() {
+        assertEquals("data_analysis:combination_query_key:＊:＊",
+                ReportBuilder.escapeWeComCode("data_analysis:combination_query_key:*:*"));
+
+        CheckReport report = baseReport();
+        KeyStructureChange key = simpleKey(
+                "data_analysis:combination_query_key:*:*",
+                "{\"a\":0}", "{\"a\":0,\"b\":0}", "b");
+        report.getKeyChanges().add(key);
+
+        String md = new ReportBuilder("[序列化结构变更]").toMarkdown(report);
+        assertTrue(md.contains("- Key --> `data_analysis:combination_query_key:＊:＊`"),
+                "通配符 * 应转成全角＊，实际:\n" + md);
+        assertFalse(md.contains("combination_query_key:*:*"),
+                "不应再输出未转义的 *:*, 实际:\n" + md);
+    }
+
     private static CheckReport baseReport() {
         CheckReport report = new CheckReport();
         report.setRepository("jnpf-java-cloud");

src/test/java/com/codechecker/cache/schema/JavaSchemaExtractorTest.java

@@ -81,4 +81,21 @@ class JavaSchemaExtractorTest {
         assertTrue(AnnotationSupport.ignoredProperties(type).contains("a"));
         assertTrue(AnnotationSupport.ignoredProperties(type).contains("b"));
     }
+
+    @Test
+    void bigDecimalMappedAsNumberNotString() {
+        String source = ""
+                + "package demo;\n"
+                + "import java.math.BigDecimal;\n"
+                + "public class MoneyVo {\n"
+                + "  private BigDecimal amount;\n"
+                + "  private String currency;\n"
+                + "}\n";
+        SourceIndex index = new SourceIndex();
+        index.addSource(source);
+        TypeSchema schema = new JavaSchemaExtractor(index, 8).extract("demo.MoneyVo");
+        assertEquals(JsonType.NUMBER, schema.getFields().get("amount").getJsonType());
+        assertEquals("BigDecimal", schema.getFields().get("amount").getJavaType());
+        assertEquals(JsonType.STRING, schema.getFields().get("currency").getJsonType());
+    }
 }

src/test/java/com/codechecker/cache/schema/SkeletonJsonRendererTest.java

@@ -80,4 +80,31 @@ class SkeletonJsonRendererTest {
                 "截断后仍应保留改动相关字段: " + truncated);
         assertEquals(true, truncated.length() >= 10);
     }
+
+    @Test
+    void numberPlaceholdersDistinguishIntegerFloatAndBigDecimal() {
+        TypeSchema schema = new TypeSchema("MoneyVo");
+        schema.add(new FieldSchema("count", JsonType.NUMBER, "Integer"));
+        schema.add(new FieldSchema("rate", JsonType.NUMBER, "Double"));
+        schema.add(new FieldSchema("amount", JsonType.NUMBER, "BigDecimal"));
+        schema.add(new FieldSchema("score", JsonType.NUMBER, "Float"));
+
+        String json = new SkeletonJsonRenderer().render(schema);
+        assertTrue(json.contains("\"count\":0"), json);
+        assertTrue(json.contains("\"rate\":0.0"), json);
+        assertTrue(json.contains("\"amount\":0.00"), json);
+        assertTrue(json.contains("\"score\":0.0"), json);
+    }
+
+    @Test
+    void integerToBigDecimalSkeletonsLookDifferent() {
+        TypeSchema oldS = new TypeSchema("A");
+        oldS.add(new FieldSchema("amount", JsonType.NUMBER, "Integer"));
+        TypeSchema newS = new TypeSchema("A");
+        newS.add(new FieldSchema("amount", JsonType.NUMBER, "BigDecimal"));
+
+        SkeletonJsonRenderer renderer = new SkeletonJsonRenderer();
+        assertEquals("{\"amount\":0}", renderer.render(oldS));
+        assertEquals("{\"amount\":0.00}", renderer.render(newS));
+    }
 }