design-readiness-check

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Design Readiness Check

设计就绪检查

Overview

概述

This skill is the quality gate between design work and implementation planning.
Its job is to answer one question clearly: is the current design complete enough to move into
writing-plans
?
It is not a replacement for
design-decision-audit
. That skill audits standalone design or plan documents. This skill checks the readiness of an in-progress design workflow before planning.
此skill是设计工作与实施规划之间的质量关卡。
它的职责是清晰回答一个问题:当前设计是否足够完备,可以进入
writing-plans
阶段?
它不能替代
design-decision-audit
。该skill用于审核独立的设计或规划文档,而本skill则是在规划前检查进行中设计工作流的就绪状态。

When to Use

使用场景

Use this skill when:
  • the design looks mostly complete
  • the next possible step is
    writing-plans
  • the user asks whether the design is ready to implement
  • the remaining concern is completeness, not discovery
Do not use this skill when:
  • the design tree is still missing
  • major branches remain vague
  • core decisions are still unresolved
  • the user is asking for a general audit of an external design document
  • design_target_type
    is missing from the design state
在以下场景中使用本skill:
  • 设计看起来基本完成
  • 下一步可能是
    writing-plans
  • 用户询问设计是否已准备好实施
  • 剩余关注点是完备性,而非探索
在以下场景中请勿使用本skill:
  • 设计树仍缺失
  • 主要分支仍模糊不清
  • 核心决策仍未解决
  • 用户要求对外部设计文档进行通用审核
  • 设计状态中缺少
    design_target_type

Workflow

工作流程

Phase A: Context Preparation

A阶段:上下文准备

Goal: Load design state and assemble context for parallel checks.
  1. Load the current
    design_state
    from the conversation context.
  2. Read the design tree content and all related context (open branches, decision nodes, risks, validation state).
  3. If
    design_state
    does not exist or the design tree is empty, return NOT READY immediately with a handoff to
    design-structure
    .
  4. If
    design_target_type
    is missing, return NOT READY immediately with a blocking issue for missing required state.
  5. Assemble a context package containing: design tree text, open branches, decision nodes, existing risks, validation entries, and
    design_target_type
    .
  6. Perform a lightweight structural integrity check using the shared rules in
    ../design-tree-core/REFERENCE.md
    :
    • mixed responsibilities inside one tree
    • parent/child ownership confusion
    • duplicated parent/child logic on the same branch
    • branches that likely should have been derived but were kept inline
目标:加载设计状态并组装上下文以进行并行检查。
  1. 从对话上下文中加载当前的
    design_state
  2. 读取设计树内容及所有相关上下文(未闭合分支、决策节点、风险、验证状态)。
  3. 如果
    design_state
    不存在或设计树为空,立即返回NOT READY,并转交至
    design-structure
  4. 如果缺少
    design_target_type
    ,立即返回NOT READY,并标记缺失必要状态的阻塞问题。
  5. 组装上下文包,包含:设计树文本、未闭合分支、决策节点、已存在的风险、验证条目以及
    design_target_type
  6. 使用
    ../design-tree-core/REFERENCE.md
    中的共享规则执行轻量级结构完整性检查:
    • 单个树中混合了不同职责
    • 父/子所有权混淆
    • 同一分支上存在重复的父/子逻辑
    • 本应派生但仍内联保留的分支

Phase B: Parallel Readiness Checks

B阶段:并行就绪检查

Goal: Run four independent checks in parallel using specialized subagents.
  1. Launch four parallel Sonnet subagents, each using one of the checker agent templates:
    • branch-checker: reads
      skills/design-readiness-check/agents/branch-checker.md
      . Fill
      {{DESIGN_TREE}}
      and
      {{CONTEXT}}
      .
    • assumption-checker: reads
      skills/design-readiness-check/agents/assumption-checker.md
      . Fill
      {{DESIGN_TREE}}
      and
      {{CONTEXT}}
      .
    • failure-checker: reads
      skills/design-readiness-check/agents/failure-checker.md
      . Fill
      {{DESIGN_TREE}}
      and
      {{CONTEXT}}
      .
    • risk-checker: reads
      skills/design-readiness-check/agents/risk-checker.md
      . Fill
      {{DESIGN_TREE}}
      and
      {{CONTEXT}}
      .
  2. Collect results from all four subagents.
Fallback: If any subagent fails or times out, the main agent performs that check inline using the same rubric from the agent template.
目标:使用专用子agent并行运行四项独立检查。
  1. 启动四个并行的Sonnet子agent,每个子agent使用一个检查器agent模板:
    • branch-checker:读取
      skills/design-readiness-check/agents/branch-checker.md
      ,填充
      {{DESIGN_TREE}}
      {{CONTEXT}}
    • assumption-checker:读取
      skills/design-readiness-check/agents/assumption-checker.md
      ,填充
      {{DESIGN_TREE}}
      {{CONTEXT}}
    • failure-checker:读取
      skills/design-readiness-check/agents/failure-checker.md
      ,填充
      {{DESIGN_TREE}}
      {{CONTEXT}}
    • risk-checker:读取
      skills/design-readiness-check/agents/risk-checker.md
      ,填充
      {{DESIGN_TREE}}
      {{CONTEXT}}
  2. 收集所有四个子agent的结果。
回退方案:如果任何子agent失败或超时,主agent将使用agent模板中的相同标准内联执行该检查。

Phase C: Verdict Synthesis

C阶段:结论综合

Goal: Combine check results into a clear readiness judgment.
  1. Map each subagent's
    status
    to a pass/fail entry in the readiness checklist:
    • required-state check → "Design target type present"
    • branch-checker → "Design tree present" and "Key branches refined"
    • assumption-checker → "Decisions resolved" (if assumptions are unresolved)
    • failure-checker → "Failure paths documented" and "Validation strategy defined"
    • risk-checker → "Blocking risks mitigated"
    • structural integrity check → "Structural integrity preserved"
  2. Build the ✓/✗ checklist diagram following Diagram Conventions.
  3. Determine verdict:
  • READY: required state is present, all four checks return
    pass
    , and no structural integrity issue is blocking
  • NOT READY: required state is missing, any check returns
    fail
    , or structural integrity issues are blocking
  1. If NOT READY, determine the handoff target based on which check failed:
    • required-state fails → hand off to
      design-structure
    • branch-checker fails → hand off to
      design-structure
      (branches missing) or
      design-refinement
      (branches weak)
    • assumption-checker fails → hand off to
      design-refinement
      (assumptions need expansion)
    • failure-checker fails → hand off to
      design-refinement
      (failure paths need adding)
    • risk-checker fails → hand off to
      decision-evaluation
      (unresolved risk decision) or
      design-refinement
      (risk documentation weak)
    • structural integrity fails → hand off to
      design-orchestrator
      (re-route ownership),
      design-structure
      (derive a child tree), or
      design-refinement
      (shrink duplicated inline logic)
  2. Update
    design_state
    with:
    • status.ready_for_planning
      : true or false
    • status.blocking_issues
      : list from failed checks
    • Updated
      open_branches
      ,
      risks
      ,
      validation
      if new information emerged
    • If a persisted design artifact exists, keep its document status aligned with the verdict:
      draft
      for NOT READY,
      ready-for-planning
      for READY
  3. Return the explicit verdict with checklist diagram. Never give a "probably ready" answer.
目标:将检查结果合并为清晰的就绪判断。
  1. 将每个子agent的
    status
    映射到就绪检查清单中的通过/未通过条目:
    • 必要状态检查 → "设计目标类型已存在"
    • branch-checker → "设计树已存在"和"关键分支已细化"
    • assumption-checker → "决策已解决"(若假设未解决)
    • failure-checker → "故障路径已记录"和"验证策略已定义"
    • risk-checker → "阻塞风险已缓解"
    • 结构完整性检查 → "结构完整性已保持"
  2. 遵循图表规范构建✓/✗检查清单图表。
  3. 确定结论:
    • READY:必要状态已存在,四项检查均返回
      pass
      ,且无阻塞性结构完整性问题
    • NOT READY:必要状态缺失,任何一项检查返回
      fail
      ,或存在阻塞性结构完整性问题
  4. 如果为NOT READY,根据哪项检查失败确定转交目标:
    • 必要状态检查失败 → 转交至
      design-structure
    • branch-checker失败 → 转交至
      design-structure
      (分支缺失)或
      design-refinement
      (分支薄弱)
    • assumption-checker失败 → 转交至
      design-refinement
      (假设需要扩展)
    • failure-checker失败 → 转交至
      design-refinement
      (需要添加故障路径)
    • risk-checker失败 → 转交至
      decision-evaluation
      (未解决的风险决策)或
      design-refinement
      (风险文档薄弱)
    • 结构完整性检查失败 → 转交至
      design-orchestrator
      (重新分配所有权)、
      design-structure
      (派生子树)或
      design-refinement
      (缩减重复的内联逻辑)
  5. 更新
    design_state
    ,包含:
    • status.ready_for_planning
      :true或false
    • status.blocking_issues
      :来自失败检查的问题列表
    • 若出现新信息,更新
      open_branches
      risks
      validation
    • 若存在持久化设计工件,使其文档状态与结论保持一致:NOT READY时为
      draft
      ,READY时为
      ready-for-planning
  6. 返回明确结论及检查清单图表。绝不能给出“可能就绪”的答案。

Readiness Standard

就绪标准

A design is ready for planning only when:
  • design_target_type
    is present
  • the main design tree is present
  • key branches are refined enough to guide implementation for the current target type
  • major decision nodes are resolved or explicitly deferred with acceptable rationale
  • failure paths and validation strategy are not missing
  • blocking risks are either mitigated or clearly acknowledged
  • tree responsibilities are not mixed in a way that breaks routing clarity
设计只有满足以下条件时才具备规划就绪状态:
  • 存在
    design_target_type
  • 主设计树已存在
  • 关键分支已细化到足以指导当前目标类型的实施
  • 主要决策节点已解决,或有可接受的理由明确推迟
  • 故障路径和验证策略无缺失
  • 阻塞风险已缓解或已明确确认
  • 树的职责未混合到破坏路由清晰度的程度

Expected Outputs

预期输出

Produce or update a
design_state
that includes:
  • design_target_type
  • open_branches
  • risks
  • validation
  • status.ready_for_planning
  • status.blocking_issues
Always return an explicit result:
  • ready for planning
  • not ready for planning
If a persisted design file is part of the workflow, its document status should also be explicit:
  • draft
    while the design is not ready
  • ready-for-planning
    after the readiness gate passes
生成或更新包含以下内容的
design_state
  • design_target_type
  • open_branches
  • risks
  • validation
  • status.ready_for_planning
  • status.blocking_issues
始终返回明确结果:
  • 已准备好进入规划阶段
  • 未准备好进入规划阶段
如果工作流中包含持久化设计文件,其文档状态也应明确:
  • 设计未就绪时为
    draft
  • 通过就绪关卡后为
    ready-for-planning

Diagram Conventions

图表规范

Present the readiness verdict as a status checklist inside a code block (no language tag):
Readiness Check
├── Design target type present   ✓
├── Design tree present          ✓
├── Key branches refined         ✓
├── Decisions resolved           ✗ (storage choice pending)
├── Failure paths documented     ✓
├── Validation strategy defined  ✓
├── Blocking risks mitigated     ✗ (migration cutover risk)
└── Structural integrity preserved ✓

Verdict: NOT READY — 2 blocking issues
Rules:
  • Use
    (pass) and
    (fail) markers with inline reason for failures
  • Max width: 78 characters
  • Always include this checklist in the readiness verdict
在代码块中(无语言标签)以状态检查清单的形式呈现就绪结论:
Readiness Check
├── Design target type present   ✓
├── Design tree present          ✓
├── Key branches refined         ✓
├── Decisions resolved           ✗ (storage choice pending)
├── Failure paths documented     ✓
├── Validation strategy defined  ✓
├── Blocking risks mitigated     ✗ (migration cutover risk)
└── Structural integrity preserved ✓

Verdict: NOT READY — 2 blocking issues
规则:
  • 使用
    (通过)和
    (未通过)标记,并为未通过项添加内联原因
  • 最大宽度:78字符
  • 就绪结论中必须包含此检查清单

Entry and Exit Criteria

进入与退出标准

Enter when:
  • the design is near completion
  • the next meaningful step may be implementation planning
Exit when:
  • you have issued a clear readiness decision
  • you have identified the next route if the design is not ready
进入条件:
  • 设计接近完成
  • 下一步可能是实施规划
退出条件:
  • 已发布明确的就绪决策
  • 已确定设计未就绪时的下一步路径

Handoff Rules

转交规则

  • Hand off to
    writing-plans
    only when the design is clearly ready.
  • Hand off to
    design-structure
    when foundational branches are still missing.
  • Hand off to
    design-refinement
    when important branches exist but are still weak.
  • Hand off to
    decision-evaluation
    when the real blocker is an unresolved decision node.
  • Hand off to
    design-orchestrator
    when the real blocker is tree ownership or routing drift rather than branch weakness.
  • Treat missing
    design_target_type
    as an immediate blocker, not as a soft warning.
  • Never give a "probably ready" answer. The result must be explicit.

  • 仅当设计明确就绪时,才转交至
    writing-plans
  • 当基础分支仍缺失时,转交至
    design-structure
  • 当重要分支存在但仍薄弱时,转交至
    design-refinement
  • 当真正的阻塞因素是未解决的决策节点时,转交至
    decision-evaluation
  • 当真正的阻塞因素是树所有权或路由偏移而非分支薄弱时,转交至
    design-orchestrator
  • 将缺失
    design_target_type
    视为立即阻塞因素,而非软警告。
  • 绝不能给出“可能就绪”的答案。结果必须明确。

Journal Integration

日志集成

When operating on a task tracked under
.agents/tasks/<task-id>/
, append a journal entry at this skill's milestone.
  • Trigger: after readiness verdict is produced
  • Reserved key(s):
    readiness
    (value must be
    ready
    or
    not-ready
    )
  • Entry shape:
    ## <ISO8601-timestamp> — design-readiness-check
    readiness: ready
    [optional ≤ 15-line body; longer content goes to artifacts/]
Resolve the task id from the explicit caller argument or
.agents/tasks/.current
. If neither resolves, skip the append; do not guess.
See
task-journal
for the full convention.
当在
.agents/tasks/<task-id>/
下跟踪的任务上操作时,在此skill的里程碑处追加日志条目。
  • 触发时机: 就绪结论生成后
  • 保留键:
    readiness
    (值必须为
    ready
    not-ready
  • 条目格式:
    ## <ISO8601-timestamp> — design-readiness-check
    readiness: ready
    [可选≤15行正文;更长内容放入artifacts/]
从显式调用者参数或
.agents/tasks/.current
解析任务ID。如果两者都无法解析,则跳过追加操作;请勿猜测。
有关完整规范,请参见
task-journal