@chenmk/superflow 0.1.1 → 0.1.3

package/assets/skills-en/superflow-pipeline/references/test-execution-template.md

@@ -1,220 +1,22 @@
-# 测试执行与结果记录模板
-
-记录测试的执行环境、执行过程和结果，作为交付物的一部分。
-
-## 文档结构
-
-```markdown
-# 测试执行报告
-
-## 1. 执行环境
-
-| 项目 | 信息 |
-|------|------|
-| 执行时间 | 2026-05-09 14:00 - 15:30 |
-| 执行人 | ____ |
-| 操作系统 | Linux / macOS / Windows |
-| JDK 版本 | 17 |
-| 数据库 | MySQL 8.0 / H2 |
-| Spring Profile | dev / test / cmk |
-| 应用端口 | 9527 |
-| 上下文路径 | / |
-
-## 2. 测试范围
-
-| 层次 | 类型 | 用例数 | 自动化 |
-|------|------|--------|--------|
-| L1 | 单元测试 | __ | [自动化] |
-| L2 | 集成测试 | __ | [自动化] |
-| L3 | 接口测试 | __ | [自动化 / 需手工] |
-| L4 | 验收测试 | __ | [需手工] |
-
-## 3. 执行方式
-
-### 3.1 自动化测试执行命令
-
-```bash
-# 全量单元测试 + 集成测试
-mvn clean test
-
-# 仅接口测试（需应用已启动）
-mvn test -Dtest=*IntegrationTest
-
-# 单测试类
-mvn test -Dtest=UserServiceTest
-```
-
-### 3.2 手工测试执行步骤
-
-1. 启动应用：`mvn spring-boot:run -Dspring-boot.run.profiles=dev`
-2. 确认健康检查：`curl http://localhost:9527/actuator/health`
-3. 按 tests.md 逐个执行 curl 命令
-4. 验证数据库状态
-5. 检查日志无 ERROR
-
-## 4. 测试结果汇总
-
-| 测试项 | 状态 | 通过 | 失败 | 跳过 | 备注 |
-|--------|------|------|------|------|------|
-| 单元测试 | ✅ / ❌ | __ | __ | __ | |
-| 集成测试 | ✅ / ❌ | __ | __ | __ | |
-| 接口测试 | ✅ / ❌ | __ | __ | __ | |
-| 回归测试 | ✅ / ❌ | __ | __ | __ | |
-
-**总体结果**: [全部通过 / 部分通过 / 未通过]
-
-## 4.1 真实测试数据与证据
-
-| 用例 ID | 真实参数 | 参数来源 | 自动化命令 | 证据等级 | 执行证据 | 结论 |
-|---------|----------|----------|------------|----------|----------|------|
-| TC-001 | parkCode=____ | 前端请求 / 用户提供 / DB查询 / Mock | curl/Newman/pytest/RestAssured | Mock only / 测试端点 / 真实入口 | 响应 + DB + 日志 | Passed / Blocked / Partial |
-
-缺少真实参数、token、站点/运营商归属、外部服务配置或设备环境时，不得填写
-`Passed`，必须标记为 `Blocked` 或 `Partial`。
-
-涉及第三方、外部集成、支付、设备、回调、客户端或跨系统链路时，测试
-Controller、mock endpoint、绕过鉴权端点只能填写 `Mock only` 或 `测试端点`。
-只有真实外部入口的 payload、响应、trace 日志、DB 证据闭环时，才能填写
-`Passed`。
-
-## 4.2 Red-Green 证据
-
-新功能、Bug 修复、CR/Px 联调补项都必须记录。RED 必须发生在生产代码修改前，
-且失败原因必须是预期业务断言，不是环境错误、编译错误、token 缺失或测试代码错误。
-
-| 用例 ID | RED 命令 | RED 失败摘要 | RED 判定 | GREEN 命令 | GREEN 通过摘要 |
-|---------|----------|---------------|----------|------------|----------------|
-| TC-001 | curl/test 命令 | 预期业务断言失败 | 正确失败/环境失败/阻塞 | 同一命令 | 响应/DB/日志均通过 |
-
-## 4.3 接口自动化证据
-
-涉及接口、CRUD、Mapper/XML、数据库字段、配置驱动、MQ、定时任务、设备、支付、
-退款、小程序或第三方链路时必须填写：
-
-| 用例 ID | Base URL | Token/Cookie 获取 | 请求命令 | 请求数据来源 | 响应断言 | DB SELECT | 日志关键词 | 结论 |
-|---------|----------|-------------------|----------|--------------|----------|-----------|------------|------|
-| TC-001 | http://localhost:____ | /captcha/image + /login | curl ... | tests.md / DB / 用户提供 | code/data/业务字段 | SELECT ... | traceId/业务日志 | Passed/Blocked/Partial |
-
-HTTP 200、字段非空、mock-only、DB-only、`BUILD SUCCESS`、`Tests are skipped`
-都不能单独作为接口通过证据。
-
-## 5. 详细结果
-
-### 5.1 通过的测试
-
-| 用例 ID | 用例名称 | 执行时间 | 验证方式 |
-|---------|----------|----------|----------|
-| TC-001 | 用户注册成功 | 2s | 接口返回 + 数据库查询 |
-
-### 5.2 失败的测试
-
-| 用例 ID | 用例名称 | 失败原因 | 处理措施 | 状态 |
-|---------|----------|----------|----------|------|
-| TC-005 | 并发下单 | 竞态条件导致超卖 | 增加分布式锁 | 已修复 ✅ |
-
-### 5.3 阻塞的测试
-
-| 用例 ID | 用例名称 | 阻塞原因 | 预计解决 |
-|---------|----------|----------|----------|
-| TC-010 | 设备回调 | 无真实设备 | Mock 验证替代 |
-
-## 6. 问题与风险
-
-| 问题 | 影响 | 应对方案 |
-|------|------|----------|
-| ____ | ____ | ____ |
-
-## 7. 数据库结构核查
-
-| 表名 | 核查命令 | 预期结构 | 实际结构 | 结论 |
-|------|----------|----------|----------|------|
-| {table_name} | `SHOW CREATE TABLE {table_name}` | {字段/索引/默认值} | {实际输出摘要} | 一致/不一致 |
-| {table_name} | `SELECT COUNT(*) FROM {table_name} WHERE {condition}` | N 条 | 实际 M 条 | 一致/不一致 |
-
-## 8. 汇总 SQL 执行记录
-
-| SQL 文件路径 | 执行的脚本（行号范围或注释标记） | 执行结果 | 复查确认 |
-|--------------|--------------------------------|----------|----------|
-| `openspec/changes/{change-id}/sql/{文件名}` | 第 XX-XX 行（如 "-- P16 业务字段补齐"） | 成功/失败 | SHOW CREATE TABLE 确认 |
-
-## 8.1 跨仓数据合同对账
-
-多仓共享表、复制 PO/Entity/Mapper、MyBatis-Plus BaseMapper 或字段迁移时必须填写：
-
-| 表 | 真源结构 | 消费仓 | 实体/Mapper/SQL 字段 | 实际库字段 | 处理结论 | 验证证据 |
-|----|----------|--------|----------------------|------------|----------|----------|
-| c_xxx | 版本总 SQL + SHOW CREATE TABLE | service-a | field_a, field_b | 一致/缺失 | 通过/阻塞/已修复 | 命令、文件、行号 |
-
-如果任一消费仓映射或查询不存在列，结论必须是 `阻塞`，不得通过给测试库补废弃字段绕过。
-
-## 9. 结论
-
-[是否达到交付标准，是否可以进入下一阶段]
-```
-
-## 任务完成 Checklist（全部满足才能标记完成）
-
-以下 checklist 用于判定一个实现任务是否真正完成。每完成一项勾选一项，**全部勾选后才能标记任务完成**。
-
-### 开发阶段（Worker 执行）
-- [ ] 已从 tests.md 提取本批次用例 ID、自动化命令和断言
-- [ ] 已在生产代码修改前执行 RED 命令
-- [ ] RED 失败原因是预期业务断言，非环境/编译/token/测试代码错误
-- [ ] RED 失败证据已回填 test-report
-- [ ] 代码已修改（git diff 确认修改了正确的文件）
-- [ ] 代码自审通过（无 CRITICAL/HIGH 问题，检查空指针/异常/事务/日志）
-- [ ] 编译通过（`mvn clean compile -DskipTests` BUILD SUCCESS）
-- [ ] 已执行同一命令确认 GREEN，并回填通过证据
-- [ ] 旧进程已停止（`pkill` + 端口确认无占用）
-- [ ] 应用已重新启动（`mvn spring-boot:run` 或 `./gradlew bootRun`）
-- [ ] 进程已验证（PID 存在、启动时间接近当前时间、端口绑定正确）
-- [ ] 健康检查通过（`curl /actuator/health` 返回 UP）
-- [ ] 日志确认是本次启动（时间戳匹配当前时间）
-
-### 接口测试阶段（Worker 逐项执行）
-- [ ] Token 已获取（完整的 `/captcha/image` + `/login` 流程，token 非空）
-- [ ] L3 接口用例已逐项执行（来自 tests.md 的每个 [自动化] 用例）
-- [ ] 每个用例的 curl 命令已记录
-- [ ] 每个用例的请求数据来源已记录
-- [ ] 每个用例的响应已记录
-- [ ] 每个用例的响应断言、DB 断言、日志断言已验证（通过/失败）
-
-### 数据库验证阶段（Worker 执行）
-- [ ] 已执行 `SHOW CREATE TABLE` 确认表结构
-- [ ] 已执行 `SELECT` 查询确认数据状态与预期一致
-- [ ] 数据库验证结果已记录
-
-### 日志检查阶段（Worker 执行）
-- [ ] 已执行 `grep ERROR` 检查应用日志
-- [ ] 无 ERROR（除已知无关错误外）
-- [ ] 关键日志已出现（如设计要求的特定标记）
-
-### 报告阶段（Worker 执行）
-- [ ] `embedded-changes/pXX-xxx/test-report.md` 已更新
-- [ ] 包含：RED/GREEN 命令和输出、curl 命令、响应摘要、数据库查询结果、日志检查结果
-- [ ] 包含：`Tests run:` 真实输出
-- [ ] 包含：进程验证证据（PID、启动时间、端口）
-
-### 独立验证阶段（Tester 执行）
-- [ ] Tester 独立执行全部用例（不参考 Worker 报告）
-- [ ] Tester 客观报告通过/失败
-
-### Leader 验证阶段（Leader 执行）
-- [ ] Leader 抽查至少 30% 关键用例
-- [ ] Leader 确认进程是新启动的
-- [ ] Leader 确认接口有真实调用证据
-
-**以上 checklist 全部勾选后，才能标记任务完成。**
-
-## 质量红线
-
-- 不允许不执行测试直接填写"通过"
-- 不允许先写生产代码再补 RED；未观察 RED 的用例不得 Passed
-- 不允许修改测试断言让失败变为通过
-- 不允许隐瞒失败测试，必须记录并说明原因
-- 不允许只用 HTTP 200、字段非空、DB-only、BUILD SUCCESS 或 skipped tests 证明业务通过
-- 不允许用占位参数或 mock-only 证据证明真实前端/外部系统链路已通过
-- 不允许用测试 Controller、mock endpoint、绕过鉴权端点证明真实外部入口已通过
-- 不允许把泛化外部失败当作通过，除非需求明确要求该失败结果
-- 不允许只检查主仓数据库结构就关闭跨仓共享表任务；必须覆盖全部消费仓实体/Mapper/SQL
-- **不允许"编译成功 + 单元测试通过"就标记任务完成，必须完成上述 checklist 全部项**
+# Test Execution Template
+
+## Test Case
+
+- ID:
+- Level: L0 unit, L1 component, L2 integration, L3 API, L4 real-entry
+- Preconditions:
+- Command:
+- RED expected failure:
+- GREEN expected pass:
+- Response assertion:
+- DB assertion:
+- Log assertion:
+- Evidence location:
+
+## Report Entry
+
+- Environment:
+- Command executed:
+- Result:
+- Evidence:
+- Blocker if not executed:

package/assets/skills-en/superflow-pipeline/references/test-guide.md

@@ -1,30 +1,16 @@
-# 测试失败修复指南
+# Test Guide
 
-## 失败类型与修复方向
+## Levels
 
-| 失败现象 | 可能原因 | 修复方向 |
-|---|---|---|
-| 应用启动失败 | 配置文件错误 / 端口冲突 / Bean 初始化失败 | 检查 `application.yml`，确认端口未被占用，检查 Bean 依赖注入 |
-| 接口返回 404 | URL 路径错误 / 上下文路径未拼接 / 接口未注册 | 核对 Controller 的 `@RequestMapping` 路径，确认 context-path 已拼接 |
-| 接口返回 500 | 空指针 / SQL 异常 / 事务回滚 | 查看日志栈追踪，定位具体异常行号，检查入参是否为空 |
-| 数据库断言失败 | SQL 未生效 / 事务未提交 / 缓存未更新 | 确认方法上有 `@Transactional`，检查 `updateByPrimaryKeySelective` 返回值 |
-| 日志断言失败 | 日志级别过滤 / 异步日志未刷新 | 检查 `logback.xml` 日志级别，确认日志输出时机 |
-| 并发测试失败 | 竞态条件 / 锁未生效 / 事务隔离级别 | 检查锁的实现（数据库锁 / 分布式锁），确认事务传播行为 |
-| 灯色下发验证失败 | DMS 未连接 / 异步任务未执行 | 确认 DMS 服务可用，检查 `@Async` 线程池配置 |
-| Excel 导出验证失败 | 文件格式错误 / 字段映射错误 | 检查 EasyExcel 或 POI 配置，确认字段注解正确 |
+- L0: unit test.
+- L1: component test.
+- L2: integration with real infrastructure such as Testcontainers.
+- L3: API or message contract test with executable command.
+- L4: real external entry such as callback, device, scheduler, client, or
+  third-party sandbox.
 
-## 修复流程
+## Rule
 
-1. **复现问题**：根据测试用例的前置条件和执行步骤，手动复现失败场景
-2. **查看日志**：启动应用时添加 `--debug` 或检查 `logs/` 目录下的日志文件
-3. **定位代码**：根据日志中的异常栈追踪，找到具体的源码位置（类名 + 方法名 + 行号）
-4. **分析根因**：对照 design.md 中的设计方案，确认实现是否与设计一致
-5. **修改代码**：修复 bug，确保不引入新的问题（回归测试）
-6. **重新测试**：重新执行失败的测试用例，确认修复成功
-7. **更新文档**：如果修复涉及设计变更，同步更新 design.md 和 tasks.md
-
-## 质量红线
-
-- 不允许跳过失败的测试用例直接提交代码
-- 不允许修改测试用例的断言标准来"让测试通过"
-- 不允许在没有测试覆盖的情况下修改核心逻辑
+When tests.md requires L3/L4, test-report.md must include real command,
+response, DB/log assertion, and final state evidence. Unit tests alone do not
+close the requirement.

package/assets/skills-en/superflow-pipeline/references/traceability-matrix.md

@@ -1,106 +1,8 @@
-# 可追溯性矩阵
-
-## 目的
-
-确保从需求到设计到实现到测试的完整可追溯，避免"设计漂移"和"实现遗漏"。
-
----
-
-## 矩阵结构
-
-```markdown
 # Traceability Matrix
 
-## 需求层（spec.md）
-
-| Req ID | Requirement | Scenario | 对应 Design | 对应 Task | 对应 Test | 对应 Prompt |
-|--------|-------------|----------|-------------|-----------|-----------|-------------|
-| R1 | 主数据支持多种业务类型 | S1.1 | design.md ## 主数据 | Task 1.1 | tests.md T1 | P1 |
-| R2 | 业务创建时校验用户资格或资源充足 | S2.1 | design.md ## 创建校验 | Task 2.1 | tests.md T2 | P2 |
-| ... | ... | ... | ... | ... | ... | ... |
-
-## 设计层（design.md）
-
-| Design 章节 | 涉及文件 | 涉及方法 | 对应 Task | 对应 Prompt |
-|-------------|----------|----------|-----------|-------------|
-| ## 主数据 | BusinessRecord.java | - | Task 1.1 | P1 |
-| ## 开通校验 | OrderService.java | validateBalance() | Task 2.1 | P2 |
-| ... | ... | ... | ... | ... |
-
-## 任务层（tasks.md）
-
-| Task ID | 文件 | 方法 | 对应 Design | 对应 Prompt | 状态 |
-|---------|------|------|-------------|-------------|------|
-| Task 1.1 | BusinessRecordMapper.xml | insert | design.md ## 主数据 | P1 | 待执行 |
-| ... | ... | ... | ... | ... | ... |
-
-## Prompt 层（prompt/*.md）
-
-| Prompt | 覆盖 Task | 覆盖 Design | 依赖 Prompt | 状态 |
-|--------|-----------|-------------|-------------|------|
-| P0 | - | - | 无 | 待执行 |
-| P1 | Task 1.1-1.5 | design.md ## 数据模型 | P0 | 待执行 |
-| ... | ... | ... | ... | ... |
-```
-
----
-
-## 一致性检查规则
-
-### 规则 1：设计漂移检查
-
-**定义**：prompt 中出现了 design.md / api.md / database-contract.md 没有定义的新方案。
-
-**检查方法**：
-1. 读取每份 prompt 的"精确实现步骤"
-2. 检查每个步骤是否能在 design.md 中找到对应章节
-3. 检查 prompt 中提到的字段、接口、MQ topic 是否在设计文档中有定义
-
-**标记方式**：
-```
-⚠️ 设计漂移：prompt/p2-charge-order.md 步骤 3 提到"使用乐观锁防止并发下单"，
-   但 design.md ## 下单流程 中未提及乐观锁策略。
-   处理方式：上报 Leader 确认 / 按设计文档原方案实现
-```
-
-### 规则 2：实现遗漏检查
-
-**定义**：design.md 有章节，但没有对应的 prompt 覆盖。
-
-**检查方法**：
-1. 遍历 design.md 的所有章节
-2. 检查每个章节是否被至少一份 prompt 覆盖
-3. 检查 api.md 中的每个接口是否被至少一份 prompt 覆盖
-
-**标记方式**：
-```
-⚠️ 实现遗漏风险：design.md ## 订单快照 描述了快照表结构和写入时机，
-   但没有 prompt 覆盖该功能（P1-P4 均未提及）。
-   处理方式：补充 P2 或新增 Pn 覆盖
-```
-
-### 规则 3：测试覆盖检查
-
-**定义**：tests.md 中的用例与 prompt 的测试要求不一致。
-
-**检查方法**：
-1. 检查每份 prompt 的"测试要求"是否覆盖了 tests.md 中对应章节的全部用例
-2. 检查 prompt 中"未执行测试及原因"是否合理
-
-**标记方式**：
-```
-⚠️ 测试覆盖不足：tests.md T3.2（边界：资源刚好等于业务阈值）
-   在 prompt/p2-charge-order.md 中未提及。
-   处理方式：补充到 P2 测试要求中
-```
-
----
-
-## 自动化检查建议
-
-在生成所有 prompt 后，执行以下检查：
+| Requirement | Scenario | API/Message | Design Section | Task | Test Case | Prompt | Report Evidence | Status |
+| --- | --- | --- | --- | --- | --- | --- | --- | --- |
+| R-001 | S-001 | api.md#... | design.md#... | tasks.md#... | tests.md#... | prompt/... | test-report.md#... | Pending |
 
-1. **生成矩阵**：从 spec.md、design.md、tasks.md、tests.md、prompt/*.md 中提取对应关系
-2. **扫描漂移**：对比 prompt 内容与设计文档，标记未定义的方案
-3. **扫描遗漏**：对比设计文档与 prompt 覆盖范围，标记未覆盖的章节
-4. **输出报告**：列出所有漂移和遗漏项，供人工确认
+Every implementation prompt must be traceable back to a requirement and forward
+to report evidence.

package/assets/skills-en/superflow-pipeline/references/validation-integrity.md

@@ -1,134 +1,16 @@
-# 验收证据完整性门禁
+# Validation Integrity
 
-Use this reference whenever SDD docs, implementation prompts, or test reports
-claim that an API, integration flow, external dependency, dropdown, enum, or
-frontend payload has been tested.
+## Anti-Fake-Delivery Rules
 
-## Hard Gates
+- Do not write pass when a command was not executed.
+- Do not replace real-entry evidence with mock evidence.
+- Do not hide skipped tests inside a success summary.
+- Do not claim app verification from an old process.
+- Do not claim DB verification without SELECT, SHOW CREATE, or equivalent proof.
+- Do not claim log verification without checking current logs.
 
-- Do not mark L3/L4 tests as passed without real evidence.
-- Do not mark any Maven or Gradle test evidence as passed when the output says
-  `Tests are skipped` / `测试被跳过`. `BUILD SUCCESS` only proves the build
-  finished; it does not prove tests ran.
-- Do not collapse multiple Maven commands into one test count. Each command
-  must have its own `Tests run` summary. If a wildcard such as
-  `-Dtest='*P47*'` does not match a needed test class, record and execute a
-  separate command for that class.
-- Do not keep deleted methods, removed classes, or old algorithms in current
-  evidence tables. Old anchors may be mentioned only as explicitly deleted or
-  historical context.
-- Do not treat a generic fallback error as success unless the requirement
-  explicitly defines that fallback as the expected business result.
-- Do not use placeholder values such as `TEST_PARK_001` to prove frontend or
-  production-like integration unless the source of that value is documented.
-- Do not let mock-only evidence close a case that requires real frontend data,
-  real database state, real operator/site/park codes, real token, or an external
-  SDK/service call.
-- Do not let a test controller, mock endpoint, bypass-auth endpoint, or direct
-  service-call endpoint close a case that requires a real external entry.
-- Do not treat one repository's schema verification as complete when sibling
-  services copy the same PO/Mapper or read the same table.
-- If required test data is missing, mark the case `Blocked: missing test data`
-  and ask for or discover the data before claiming completion.
+## Allowed Outcomes
 
-## tests.md Requirements
-
-For every L3/L4 or frontend integration case, include these columns or equivalent
-fields:
-
-| Field | Required content |
-|---|---|
-| 用例 ID | Stable test id, referenced from spec/tasks/prompt |
-| 接口/流程 | Full URL or user workflow |
-| 数据来源 | User-provided, frontend captured payload, DB query, seeded fixture, or mock |
-| 关键参数 | Real values such as parkCode/operatorId/plotId/templateId; no unexplained placeholders |
-| 预期断言 | Response, DB, log, and external-call assertions |
-| 证据要求 | curl/request/response/DB/log/external response required in test-report.md |
-| 证据等级 | Mock only / Test endpoint / Real external entry |
-
-For dropdowns, enums, external SDK calls, device callbacks, payment/refund,
-invoice, export, import, and cross-system flows, add at least one real-data case
-or explicitly mark the case blocked.
-
-## test-report.md Requirements
-
-Every passed L3/L4 case must include:
-
-- test id and requirement/spec reference
-- environment, profile, port, context path, branch, and commit
-- exact command or curl
-- real request parameters and where they came from
-- raw or summarized response
-- assertion result
-- database evidence when state changes or persistence matters
-- relevant logs
-- external SDK/service request and response when external dependencies matter
-- evidence level: `Mock only`, `Test endpoint`, `Partial real entry`, or
-  `Real integration passed`
-- for cross-repository database contracts: source schema, each consumer
-  repository, entity/Mapper/SQL fields, real database columns, and conclusion
-- exact command-to-result mapping for every automated test command. If the same
-  command appears with different `Tests run` values, the report is not
-  acceptable until reconciled.
-- for sibling repositories whose build config skips tests by default, record the
-  direct command result (`Tests are skipped` if applicable), the temporary
-  change or profile that makes tests run, the real `Tests run` output, and the
-  clean git status after rollback.
-
-If any of those are absent, status must be `Not executed`, `Blocked`, or
-`Partially verified`, not `Passed`.
-
-Before delivery, run:
-
-```bash
-~/.codex/hooks/superflow-test-report-lint.py <embedded test-report.md>
-```
-
-For root aggregate reports, run the same script with `--warn-only` and review
-all warnings before signing off.
-
-## Cross-Repository Schema Rule
-
-When a table is shared across repositories or a service copies another
-service's PO/Entity/Mapper:
-
-- identify the source-of-truth schema: version SQL, database contract, and
-  `SHOW CREATE TABLE`
-- list every consuming repository and runtime path, including interconnect,
-  callback, scheduled job, import/export, and test endpoint paths
-- compare MyBatis-Plus `@TableName` entity fields, `BaseMapper` default
-  selection, Mapper XML/resultMap, and handwritten SQL against real database
-  columns
-- fields removed from the final schema must be removed from mapped entities or
-  marked `@TableField(exist = false)`
-- if code still queries a removed column, the fix is code/schema contract
-  alignment, not adding obsolete columns back to test database
-
-The case is blocked if any consumer repository maps or queries a non-existent
-column.
-
-## Red-Green Rule
-
-For bug fixes and CR/Px follow-ups:
-
-1. Record the red evidence first: the exact failing test, curl, or workflow
-   before the fix, including the error response/log/DB mismatch.
-2. Apply the fix.
-3. Run the same evidence path again and record the green result.
-
-For new features where no old failing path exists, use a negative control:
-
-- one invalid or unauthorized case fails with the expected error
-- one real happy-path case passes with response and DB/log evidence
-
-## External Dependency Rule
-
-When an external service is unavailable:
-
-- record the external request parameters, endpoint/config source, exception type,
-  external code/message if available, and whether the failure is environmental or
-  code-related
-- do not mark the business case passed unless the requirement only promises a
-  stable error under external failure
-- keep the delivery conclusion explicit: `Blocked by external dependency`,
-  `Code path verified with mock`, or `Real integration passed`
+- Passed: evidence exists and matches tests.md.
+- Partially verified: some evidence exists, and gaps are explicit.
+- Blocked: external dependency or environment prevents execution, with reason.

package/assets/skills-en/superflow-pipeline/scripts/superflow-guard.sh

@@ -244,13 +244,13 @@ require_hash_recorded() {
 
 change_has_field_status_risk() {
   grep -RIEiq \
-    '字段值|状态|枚举|online|offline|上线|下线|删除|恢复|同步标记|sync marker|payment|refund|支付|退款|第三方状态|running_status|offline_time' \
+    'field value|status|enum|online|offline|delete|restore|sync marker|payment|refund|third-party state|running_status|offline_time' \
     "$CHANGE_DIR"/*.md "$CHANGE_DIR"/**/*.md 2>/dev/null
 }
 
 change_has_architecture_boundary_risk() {
   grep -RIEiq \
-    '跨仓|跨服务|sibling|SDK|MQ|topic|consumer|scheduler|定时|device|设备|callback|回调|third[- ]party|第三方|mini[- ]program|小程序|gateway|网关|adapter|适配|protocol|协议|interconnect|互联互通|调用链|入口|出口' \
+    'cross-repo|cross-service|sibling|SDK|MQ|topic|consumer|scheduler|scheduled|device|callback|third[- ]party|mini[- ]program|gateway|adapter|protocol|interconnect|call chain|entry|exit' \
     "$CHANGE_DIR"/*.md "$CHANGE_DIR"/**/*.md 2>/dev/null
 }
 
@@ -304,12 +304,12 @@ case "$PHASE" in
     require_handoff_current
     require_beta_handoff_structure
     require_hash_recorded
-    require_grep 'Superpowers Technical Design Handoff|Superpower 技术详设|technical_design|源码级 HOW' design.md "Superpowers technical design handoff"
-    require_grep 'OpenSpec/SDD|WHAT|合同|API|DB|tests|验收' design.md "OpenSpec/SDD contract boundary"
-    require_grep '文档完整性|完整文档|proposal.md|api.md|design.md|tasks.md|tests.md' sdd-quality-gate.md "document completeness gate"
-    require_grep 'RED|红绿|GREEN' tests.md "RED/GREEN test contract"
-    require_grep 'curl|Postman|Newman|pytest|RestAssured|自动化命令' tests.md "interface automation command"
-    require_grep 'handoff_hash|sdd-context|上下文包|防漂移' sdd-quality-gate.md "handoff/context-drift gate"
+    require_grep 'Superpowers Technical Design Handoff|Superpowers Technical Design|technical_design|source-level HOW' design.md "Superpowers technical design handoff"
+    require_grep 'OpenSpec/SDD|WHAT|contract|API|DB|tests|acceptance' design.md "OpenSpec/SDD contract boundary"
+    require_grep 'document completeness|complete documents|proposal.md|api.md|design.md|tasks.md|tests.md' sdd-quality-gate.md "document completeness gate"
+    require_grep 'RED|red-green|GREEN' tests.md "RED/GREEN test contract"
+    require_grep 'curl|Postman|Newman|pytest|RestAssured|automation command' tests.md "interface automation command"
+    require_grep 'handoff_hash|sdd-context|context package|anti-drift' sdd-quality-gate.md "handoff/context-drift gate"
     require_markdown_links_valid
     transition_event="docs-complete"
     ;;
@@ -334,18 +334,18 @@ case "$PHASE" in
       require_optional_state_path technical_design "technical_design"
       technical_design_rel="$(state_get technical_design)"
       if [[ -n "${technical_design_rel:-}" && "$technical_design_rel" != "null" ]]; then
-        require_grep 'Superpowers Technical Design|Superpower 技术详设|Technical Design|源码级 HOW' "$technical_design_rel" "Superpowers technical design"
-        require_grep 'OpenSpec/SDD|canonical|事实源|不得覆盖|不能覆盖' "$technical_design_rel" "technical design canonical boundary"
+        require_grep 'Superpowers Technical Design|Superpowers Technical Design|Technical Design|source-level HOW' "$technical_design_rel" "Superpowers technical design"
+        require_grep 'OpenSpec/SDD|canonical|source of truth|must not override|must not override' "$technical_design_rel" "technical design canonical boundary"
         if change_has_architecture_boundary_risk; then
-          require_grep 'Architecture Boundary And Call Direction|架构边界与调用方向|模块职责.*调用方向|owner.*入口.*出口|禁止绕路' "$technical_design_rel" "architecture boundary and call direction matrix"
+          require_grep 'Architecture Boundary And Call Direction|Architecture Boundary And Call Direction|module responsibility.*call direction|owner.*entry.*exit|no bypass' "$technical_design_rel" "architecture boundary and call direction matrix"
         fi
         if change_has_field_status_risk; then
-          require_grep 'Field And Status Reverse Impact|字段/状态反向影响面|写入点.*读取|读取/过滤点|派生/同步点' "$technical_design_rel" "field/status reverse impact matrix"
+          require_grep 'Field And Status Reverse Impact|Field And Status Reverse Impact|write point.*read|read/filter points|derived/sync points' "$technical_design_rel" "field/status reverse impact matrix"
         fi
       fi
     fi
-    require_grep 'Superpowers Technical Design Handoff|Superpower 技术详设|technical_design|源码级 HOW' design.md "Superpowers technical design handoff"
-    require_grep 'technical_design|Superpowers 技术详设|源码级 HOW' sdd-quality-gate.md "technical design quality gate"
+    require_grep 'Superpowers Technical Design Handoff|Superpowers Technical Design|technical_design|source-level HOW' design.md "Superpowers technical design handoff"
+    require_grep 'technical_design|Superpowers Technical Design|source-level HOW' sdd-quality-gate.md "technical design quality gate"
     require_markdown_links_valid
     transition_event="design-complete"
     ;;
@@ -357,18 +357,18 @@ case "$PHASE" in
     require_hash_recorded
     require_any_prompt
     require_prompt_set
-    require_grep 'OpenSpec/SDD|设计事实源|canonical|事实源' design.md "canonical source boundary"
+    require_grep 'OpenSpec/SDD|canonical design source|canonical|source of truth' design.md "canonical source boundary"
     prompt_rel="$(first_prompt_rel)"
     if [[ -n "${prompt_rel:-}" ]]; then
-      require_grep 'Superpower 技术详设继承|Superpower 执行策略继承|technical_design|源码级 HOW' "$prompt_rel" "prompt Superpower strategy inheritance"
+      require_grep 'Superpowers Technical Design inheritance|Superpowers execution strategy inheritance|technical_design|source-level HOW' "$prompt_rel" "prompt Superpower strategy inheritance"
       technical_design_rel="$(state_get technical_design)"
       if [[ -n "${technical_design_rel:-}" && "$technical_design_rel" != "null" ]]; then
-        require_grep 'Superpower 技术详设继承|technical_design|源码级 HOW|Technical Design' "$prompt_rel" "prompt Superpowers technical design inheritance"
+        require_grep 'Superpowers Technical Design inheritance|technical_design|source-level HOW|Technical Design' "$prompt_rel" "prompt Superpowers technical design inheritance"
         if change_has_field_status_risk; then
-          require_grep '字段/状态反向影响面|Field And Status Reverse Impact|读取/过滤点|派生/同步点' "$prompt_rel" "prompt field/status reverse impact inheritance"
+          require_grep 'Field And Status Reverse Impact|Field And Status Reverse Impact|read/filter points|derived/sync points' "$prompt_rel" "prompt field/status reverse impact inheritance"
         fi
       fi
-      require_grep '上下文防漂移|handoff_hash|sdd-context' "$prompt_rel" "prompt context drift inheritance"
+      require_grep 'context anti-drift|handoff_hash|sdd-context' "$prompt_rel" "prompt context drift inheritance"
     fi
     require_state_value build_mode "build_mode"
     require_state_value isolation "isolation"
@@ -387,10 +387,10 @@ case "$PHASE" in
   verify)
     require_state_file
     require_file test-report.md
-    require_grep 'RED|失败证据' test-report.md "RED evidence"
-    require_grep 'GREEN|通过证据' test-report.md "GREEN evidence"
-    require_grep '接口自动化|curl|Postman|Newman|pytest|RestAssured' test-report.md "interface automation evidence"
-    require_grep 'DB|数据库|SELECT|SHOW CREATE' test-report.md "DB evidence"
+    require_grep 'RED|failure evidence' test-report.md "RED evidence"
+    require_grep 'GREEN|pass evidence' test-report.md "GREEN evidence"
+    require_grep 'interface automation|curl|Postman|Newman|pytest|RestAssured' test-report.md "interface automation evidence"
+    require_grep 'DB|database|SELECT|SHOW CREATE' test-report.md "DB evidence"
     require_grep 'superflow-verify-integration|superflow-delivery-check|superflow-test-report-lint' test-report.md "SuperBridge Flow hook/script evidence"
     require_test_report_lint
     require_markdown_links_valid

package/assets/skills-en/superflow-pipeline/scripts/superflow-state.sh

@@ -494,7 +494,7 @@ cmd_scale() {
     changed_files="$(git -C "$CHANGE_DIR" diff --name-only HEAD 2>/dev/null | wc -l | tr -d ' ')"
   fi
 
-  risk_hits="$(grep -REi 'api|接口|curl|SQL|数据库|Mapper|XML|跨仓|状态字段|枚举|真实入口|payment|refund|MQ|定时任务|第三方|SDK' "$CHANGE_DIR"/*.md "$CHANGE_DIR"/**/*.md 2>/dev/null | wc -l | tr -d ' ')"
+  risk_hits="$(grep -REi 'api|api|curl|SQL|database|Mapper|XML|cross-repo|status field|enum|real entry|payment|refund|MQ|scheduled job|third party|SDK' "$CHANGE_DIR"/*.md "$CHANGE_DIR"/**/*.md 2>/dev/null | wc -l | tr -d ' ')"
 
   mode="light"
   if [[ "$task_count" -gt 3 || "$capability_count" -gt 1 || "$changed_files" -gt 4 || "$risk_hits" -gt 0 ]]; then