spec-test

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Spec Pack 验证(Verification)主技能:spec-test

Spec Pack Verification Main Skill: spec-test

本技能是 verification 阶段的 worker skill 入口:负责统一门禁、输入/输出约束、追溯与 DoD 自检。实际产物由子技能
spec-test-*
生成。
权威口径:
design/aisdlc_spec_verification.md
(本文档即本技能的 SSOT)。

This skill is the worker skill entry for the verification phase: it is responsible for unified access control, input/output constraints, traceability and DoD self-check. Actual artifacts are generated by the sub-skill
spec-test-*
.
Authoritative Reference:
design/aisdlc_spec_verification.md
(this document is the SSOT for this skill).

何时使用

When to Use

  • 需要为某个 Spec Pack 产出或更新以下任一文件:
    • verification/test-plan.md
    • verification/usecase.md
    • verification/suites.md
    • verification/report-{date}-{version}.md
  • 你发现自己开始出现以下“危险信号”(说明很容易违规):
    • 想“先随便写到
      ./verification/
      ,之后再迁移”
    • 想跳过
      requirements/*
      或项目级 memory,说“先写个模板占位”
    • 想用
      TBD/待补
      代替 AC 追溯与影响面
    • 想在技能内决定下一步流程(越权路由)
  • You need to produce or update any of the following files for a Spec Pack:
    • verification/test-plan.md
    • verification/usecase.md
    • verification/suites.md
    • verification/report-{date}-{version}.md
  • You start to notice the following "red flags" (indicating high risk of rule violation):
    • You want to "just write to
      ./verification/
      first and migrate later"
    • You want to skip
      requirements/*
      or project-level memory, saying "just write a template placeholder first"
    • You want to use
      TBD/to be filled
      instead of AC traceability and impact scope
    • You want to decide the next process within the skill (unauthorized routing)

何时不使用

When Not to Use

  • 你在做“下一步该走哪里/是否跳过 verification”的分流判断:这属于 Router,应该回到
    using-aisdlc
  • 你要写自动化测试代码:verification 阶段此技能只负责文档产物与门禁,不要求自动化代码实现。

  • You are making the diversion judgment of "where to go next/whether to skip verification": this belongs to the Router, you should return to
    using-aisdlc
    .
  • You need to write automated test code: this skill in the verification phase is only responsible for document artifacts and access control, and does not require automated code implementation.

硬门禁(必须遵守)

Hard Access Control (Must Comply)

1) 先定位再读写:必须先得到
FEATURE_DIR

1) Locate before reading/writing:
FEATURE_DIR
must be obtained first

  • 凡读写
    verification/*
    ,必须先运行
    spec-context
    获取
    FEATURE_DIR
    并回显
  • spec-context
    失败即停止
    :不得猜路径、不得用当前工作目录推断、不得“先写到根目录再说”。
  • For any read/write operation on
    verification/*
    , you must run
    spec-context
    first to get
    FEATURE_DIR
    and echo it
    .
  • Stop immediately if
    spec-context
    fails
    : do not guess the path, do not infer from the current working directory, do not "write to the root directory first".

2) 必读上下文(渐进式披露)

2) Must-read context (progressive disclosure)

读取失败或不存在时,必须显式写出
CONTEXT GAP
,不得静默跳过:
  • project/memory/product.md
  • project/memory/tech.md
  • project/memory/glossary.md
需求级最小必要输入(至少其一必须存在):
  • {FEATURE_DIR}/requirements/solution.md
    {FEATURE_DIR}/requirements/prd.md
If reading fails or the context does not exist, you must explicitly write
CONTEXT GAP
, and do not skip silently:
  • project/memory/product.md
  • project/memory/tech.md
  • project/memory/glossary.md
Minimum required input at requirement level (at least one must exist):
  • {FEATURE_DIR}/requirements/solution.md
    or
    {FEATURE_DIR}/requirements/prd.md

3) Worker 边界:禁止越权路由

3) Worker boundary: Unauthorized routing is prohibited

  • 本技能与子技能 只负责门禁 + 产物落盘 + DoD 自检
  • 完成后 不得在技能内部决定“下一步做什么”,只能输出
    ROUTER_SUMMARY
    ,并提示“回到 Router(using-aisdlc)继续路由”。

  • This skill and its sub-skills are only responsible for access control + artifact storage + DoD self-check.
  • After completion, you must not decide "what to do next" inside the skill, you can only output
    ROUTER_SUMMARY
    and prompt "return to Router (using-aisdlc) to continue routing".

子技能一览(调用/路由入口)

Sub-skill List (Call/Routing Entry)

子技能产物(落盘)目标
spec-test-plan
{FEATURE_DIR}/verification/test-plan.md
冻结范围/策略/环境/准入准出/风险清单
spec-test-usecase
{FEATURE_DIR}/verification/usecase.md
把 AC 转为可执行、可生成自动化的用例结构
spec-test-suites
{FEATURE_DIR}/verification/suites.md
组织 smoke/regression/targeted 套件与执行顺序
spec-test-execute
{FEATURE_DIR}/verification/report-{date}-{version}.md
产出结论性报告:覆盖/结果/阻断/缺陷引用/风险建议

Sub-skillArtifacts (Persistent Storage)Purpose
spec-test-plan
{FEATURE_DIR}/verification/test-plan.md
Freeze scope/strategy/environment/entry & exit criteria/risk list
spec-test-usecase
{FEATURE_DIR}/verification/usecase.md
Convert AC into executable, automation-ready test case structure
spec-test-suites
{FEATURE_DIR}/verification/suites.md
Organize smoke/regression/targeted suites and execution order
spec-test-execute
{FEATURE_DIR}/verification/report-{date}-{version}.md
Produce conclusive report: coverage/results/blockers/defect references/risk suggestions

反模式与显式反制(来自基线压力场景)

Anti-patterns and Explicit Countermeasures (From Baseline Pressure Scenarios)

常见借口/压力话术现实与反制
“赶时间,先写到
./verification/
,后面再搬”
禁止猜路径。 未定位
FEATURE_DIR
就落盘会制造不可追溯的垃圾产物;必须先
spec-context
“requirements 太慢,先写模板占位”verification 的核心是 AC → 用例/报告追溯;不读
solution/prd
等于放弃可验证口径。允许
CONTEXT GAP
,但不允许用
TBD
伪装完成。
“负责人说别管门禁,能用就行”worker skill 的价值就是门禁;如果必须偏离流程,只能在产物中写明 风险与后果,并把偏离点升级为显式
CONTEXT GAP / PROCESS DEVIATION
,且仍不得猜路径。
“我顺手告诉你下一步该做什么”禁止越权路由。 只能输出
ROUTER_SUMMARY
,下一步由
using-aisdlc
决策。

Common Excuses/Pressure StatementsFacts and Countermeasures
"We are in a hurry, just write to
./verification/
first and move it later"
Path guessing is prohibited. Storing artifacts without locating
FEATURE_DIR
will create untraceable junk artifacts; you must run
spec-context
first.
"Requirements are coming too slow, just write a template placeholder first"The core of verification is AC → test case/report traceability; not reading
solution/prd
equals giving up verifiable criteria.
CONTEXT GAP
is allowed, but using
TBD
to pretend completion is not allowed.
"The person in charge said ignore the access control, as long as it works"The value of worker skill is access control; if you have to deviate from the process, you can only write down the risks and consequences in the artifact, upgrade the deviation to an explicit
CONTEXT GAP / PROCESS DEVIATION
, and still cannot guess the path.
"I'll tell you what to do next by the way"Unauthorized routing is prohibited. You can only output
ROUTER_SUMMARY
, and the next step is decided by
using-aisdlc
.

输出约定(统一给 Router 消费)

Output Convention (Unified for Router Consumption)

任一
spec-test
/
spec-test-*
完成时,必须执行「完成后输出与自动路由」三步骤(与
spec-product-clarify
等 worker skill 一致):
  1. 输出 ROUTER_SUMMARY(YAML 形态,供 Router 决策;具体 stage/artifacts 以各子技能为准)
  2. 立即执行
    using-aisdlc
    :将
    ROUTER_SUMMARY
    作为路由输入传递,由 Router 判定下一步并自动推进(无需等待用户说「继续」)
  3. 对话输出:在调用 using-aisdlc 前,可简短说明「本阶段产物已落盘,正在调用 using-aisdlc 路由下一步。」
兼容说明:历史上某些文档会使用自由文本/自定义字段形式的“router summary”。在本仓库 Spec Pack 流程中,以
using-aisdlc
定义的 YAML
ROUTER_SUMMARY
为唯一权威输出格式。
When any
spec-test
/
spec-test-*
is completed, you must perform the three steps of "post-completion output and automatic routing" (consistent with worker skills such as
spec-product-clarify
):
  1. Output ROUTER_SUMMARY (in YAML format for Router decision; specific stage/artifacts are subject to each sub-skill)
  2. Execute
    using-aisdlc
    immediately
    : Pass
    ROUTER_SUMMARY
    as routing input, and the Router will determine the next step and automatically proceed (no need to wait for the user to say "continue")
  3. Conversation output: Before calling using-aisdlc, you can briefly state "The artifacts of this phase have been stored, and using-aisdlc is being called to route the next step."
Compatibility note: Historically, some documents used free text/custom field form of "router summary". In the Spec Pack process of this repository, the YAML
ROUTER_SUMMARY
defined by
using-aisdlc
is the only authoritative output format.