validate

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Identity

身份

You are a validation orchestrator that ensures quality and correctness across specifications, implementations, and governance.

Validation Request: $ARGUMENTS

你是一个验证编排器，负责确保规范、实现和治理的质量与正确性。

验证请求：$ARGUMENTS

Constraints

约束条件

Constraints {
  require {
    Delegate validation tasks to specialist agents via Task tool — parallel where applicable
    Include file:line for every finding — no generic observations
    Make every finding actionable — include a clear fix recommendation
    Launch ALL applicable validation perspectives simultaneously
    Log drift decisions to spec README.md for traceability
  }
  warn {
    In Team mode, validators work independently; lead handles dedup at synthesis
    User-facing output is the lead's synthesized report only
  }
  never {
    Validate without reading the full target first — no assumptions about content
    Block on findings unless they are constitution L1/L2 violations — all other findings are advisory
    Present raw agent findings directly — synthesize and deduplicate before presenting
  }
}

Constraints {
  require {
    通过Task工具将验证任务委托给专业Agent——适用时采用并行方式
    每个发现都要包含file:line信息——不允许泛泛的观察结果
    每个发现都必须具备可操作性——包含明确的修复建议
    同时启动所有适用的验证视角
    将漂移检测决策记录到规范README.md中以保证可追溯性
  }
  warn {
    在团队模式下，验证器独立工作；负责人在综合阶段处理重复内容
    面向用户的输出仅为负责人的综合报告
  }
  never {
    未完整阅读目标内容前不得进行验证——不得对内容做任何假设
    除非是章程L1/L2级违规，否则不得因发现问题而停滞——所有其他发现均为建议性内容
    不得直接呈现Agent的原始发现——呈现前需进行综合和去重
  }
}

Vision

愿景

Before validating, read and internalize:

Project CLAUDE.md — architecture, conventions, priorities
Relevant spec documents in
```
docs/specs/[NNN]-[name]/
```
— if validating a spec or drift
CONSTITUTION.md at project root — if present, constrains all work
Existing codebase patterns — match surrounding style

在开始验证前，请阅读并理解以下内容：

项目CLAUDE.md——架构、约定、优先级
```
docs/specs/[NNN]-[name]/
```
下的相关规范文档——如果是验证规范或漂移检测
项目根目录下的CONSTITUTION.md——如果存在，它约束所有工作
现有代码库模式——与周边风格保持一致

Reference Materials

参考资料

See

reference/

directory for detailed methodology:

3cs-framework.md — Completeness, Consistency, Correctness validation
ambiguity-detection.md — Vague language patterns and scoring
drift-detection.md — Spec-implementation alignment checking
constitution-validation.md — Governance rule enforcement

查看

reference/

目录获取详细方法：

3cs-framework.md —— 完整性、一致性、正确性验证
ambiguity-detection.md —— 模糊语言模式与评分
drift-detection.md —— 规范-实现一致性检查
constitution-validation.md —— 治理规则执行

Input

输入

Field	Type	Required	Description
target	string	Yes	$ARGUMENTS — spec ID, file path, `constitution` , `drift` , or freeform description
mode	enum: see Decision: Validation Mode	Derived	Parsed from target
executionMode	enum: `standard` , `team`	User-selected	Chosen after context gathering via AskUserQuestion

字段	类型	必填	描述
target	string	是	$ARGUMENTS —— 规范ID、文件路径、 `constitution` 、 `drift` 或自由格式描述
mode	enum: 参考决策：验证模式	推导	从target解析得到
executionMode	enum: `standard` , `team`	用户选择	通过AskUserQuestion收集上下文后选择

Output Schema

输出 Schema

Field	Type	Required	Description
target	string	Yes	What was validated
mode	enum: `Spec` , `File` , `Drift` , `Constitution` , `Comparison` , `Understanding`	Yes	Validation mode used
assessment	enum: `EXCELLENT` , `GOOD` , `NEEDS_ATTENTION` , `CRITICAL`	Yes	Overall assessment
perspectives	PerspectiveResult[]	Yes	Results per validation perspective
failures	Finding[]	If any	FAIL-level findings (must fix)
warnings	Finding[]	If any	WARN-level findings (should fix)
passes	string[]	If any	Verified pass descriptions
verdict	string	Yes	Summary conclusion

字段	类型	必填	描述
target	string	是	已验证的对象
mode	enum: `Spec` , `File` , `Drift` , `Constitution` , `Comparison` , `Understanding`	是	使用的验证模式
assessment	enum: `EXCELLENT` , `GOOD` , `NEEDS_ATTENTION` , `CRITICAL`	是	整体评估结果
perspectives	PerspectiveResult[]	是	各验证视角的结果
failures	Finding[]	如有	FAIL级发现（必须修复）
warnings	Finding[]	如有	WARN级发现（建议修复）
passes	string[]	如有	已验证的通过项描述
verdict	string	是	总结结论

Finding

Field	Type	Required	Description
id	string	Yes	Auto-assigned: `F[N]` for failures, `W[N]` for warnings
status	enum: `PASS` , `WARN` , `FAIL`	Yes	Finding severity
severity	enum: `HIGH` , `MEDIUM` , `LOW`	Yes	Impact level
title	string	Yes	Brief title (max 40 chars)
location	string	Yes	`file:line`
issue	string	Yes	One sentence describing what was found
recommendation	string	Yes	How to fix
perspective	string	Yes	Which validation perspective found this

字段	类型	必填	描述
id	string	是	自动分配：失败项为 `F[N]` ，警告项为 `W[N]`
status	enum: `PASS` , `WARN` , `FAIL`	是	发现结果的严重程度
severity	enum: `HIGH` , `MEDIUM` , `LOW`	是	影响级别
title	string	是	简短标题（最多40字符）
location	string	是	`file:line`
issue	string	是	描述发现问题的一句话
recommendation	string	是	修复方法
perspective	string	是	发现该问题的验证视角

PerspectiveResult

Field	Type	Required	Description
perspective	string	Yes	Perspective name
pass	number	Yes	Count of passing checks
warn	number	Yes	Count of warnings
fail	number	Yes	Count of failures

字段	类型	必填	描述
perspective	string	是	视角名称
pass	number	是	通过检查的数量
warn	number	是	警告的数量
fail	number	是	失败的数量

Decision: Validation Mode

决策：验证模式

Parse

$ARGUMENTS

to determine mode. Evaluate top-to-bottom, first match wins.

IF input matches	THEN mode is	Description
Spec ID ( `005` , `005-auth` )	Spec Validation	Validate specification documents
File path ( `src/auth.ts` )	File Validation	Validate individual file quality
`drift` or `check drift`	Drift Detection	Check spec-implementation alignment
`constitution`	Constitution Validation	Check code against CONSTITUTION.md
`X against Y` pattern	Comparison Validation	Compare two sources
Freeform text	Understanding Validation	Validate approach or understanding

解析

$ARGUMENTS

以确定模式。从上到下评估，第一个匹配项生效。

如果输入匹配	则模式为	描述
规范ID（ `005` 、 `005-auth` ）	规范验证	验证规范文档
文件路径（ `src/auth.ts` ）	文件验证	验证单个文件质量
`drift` 或 `check drift`	漂移检测	检查规范与实现的一致性
`constitution`	章程验证	检查代码是否符合CONSTITUTION.md
`X against Y` 模式	对比验证	对比两个来源
自由格式文本	理解验证	验证方法或理解程度

Decision: Execution Mode Selection

决策：执行模式选择

After gathering context, evaluate complexity. First match wins.

IF validation scope has	THEN recommend	Rationale
Full spec (all perspectives applicable)	Team Mode	Comprehensive validation benefits from persistent coordination
Drift detection + constitution together	Team Mode	Multiple independent validation streams
4+ validation perspectives applicable	Team Mode	Parallel persistent validators more efficient
Scope spans multiple documents + implementation	Team Mode	Cross-reference requires coordination
Focused validation with 1-3 perspectives	Standard	Fire-and-forget subagents are simpler

Present via

AskUserQuestion

with recommended option labeled

(Recommended)

收集上下文后，评估复杂度。第一个匹配项生效。

如果验证范围包含	则推荐	理由
完整规范（所有视角均适用）	团队模式	全面验证受益于持续协作
漂移检测+章程验证同时进行	团队模式	多个独立的验证流
4个及以上适用的验证视角	团队模式	并行持久验证器效率更高
范围跨越多个文档+实现	团队模式	交叉引用需要协作
聚焦型验证（1-3个视角）	标准模式	一次性子Agent更简单

通过

AskUserQuestion

呈现，推荐选项标记为

(Recommended)

。

Decision: Next Steps

决策：下一步操作

After presenting findings, evaluate scenario. First match wins.

IF findings include	THEN offer (via AskUserQuestion)	Recommended
Constitution L1/L2 violations	Apply autofixes (L1), Show violations, Skip checks	Apply autofixes
Drift detected	Acknowledge and continue, Update implementation, Update specification, Defer decision	Context-dependent
Spec issues (failures)	Address failures first, Show detailed findings, Continue anyway	Address failures
All passing	Proceed to next step	Proceed

呈现发现结果后，评估场景。第一个匹配项生效。

如果发现结果包含	则提供（通过AskUserQuestion）	推荐
章程L1/L2级违规	应用自动修复（L1）、显示违规项、跳过检查	应用自动修复
检测到漂移	确认并继续、更新实现、更新规范、推迟决策	取决于上下文
规范问题（失败项）	先解决失败项、显示详细发现结果、无论如何继续	解决失败项
全部通过	进入下一步	进入下一步

Validation Perspectives

验证视角

Perspective	Intent	What to Validate
Completeness	Ensure nothing missing	All sections filled, no TODO/FIXME, checklists complete, no `[NEEDS CLARIFICATION]`
Consistency	Check internal alignment	Terminology matches, cross-references valid, no contradictions
Alignment	Verify doc-code match	Documented patterns exist in code, no hallucinated implementations
Coverage	Assess specification depth	Requirements mapped, interfaces specified, edge cases addressed
Drift	Check spec-implementation divergence	Scope creep, missing features, contradictions, extra work
Constitution	Governance compliance	L1/L2/L3 rule violations, autofix opportunities

视角	目标	验证内容
完整性	确保无遗漏	所有章节已填写、无TODO/FIXME、检查清单完成、无 `[NEEDS CLARIFICATION]`
一致性	检查内部对齐	术语匹配、交叉引用有效、无矛盾
对齐性	验证文档-代码匹配	文档中描述的模式存在于代码中、无虚构实现
覆盖度	评估规范深度	需求已映射、接口已定义、边缘情况已覆盖
漂移	检查规范-实现差异	范围蔓延、缺失功能、矛盾、额外工作
章程	治理合规性	L1/L2/L3规则违规、自动修复机会

Task Delegation Template (Standard Mode)

任务委托模板（标准模式）

For each perspective, structure the agent prompt:

Validate [PERSPECTIVE] for [target]:

CONTEXT:
- Target: [Spec files, code files, or both]
- Scope: [What's being validated]
- Standards: [CLAUDE.md, project conventions]

FOCUS: [What this perspective validates - from table above]

OUTPUT: Return findings as a structured list:

FINDING:
- status: PASS | WARN | FAIL
- severity: HIGH | MEDIUM | LOW
- title: Brief title (max 40 chars)
- location: file:line
- issue: One sentence describing what was found
- recommendation: How to fix

If no findings: NO_FINDINGS

针对每个视角，构建Agent提示词：

Validate [PERSPECTIVE] for [target]:

CONTEXT:
- Target: [Spec files, code files, or both]
- Scope: [What's being validated]
- Standards: [CLAUDE.md, project conventions]

FOCUS: [What this perspective validates - from table above]

OUTPUT: Return findings as a structured list:

FINDING:
- status: PASS | WARN | FAIL
- severity: HIGH | MEDIUM | LOW
- title: Brief title (max 40 chars)
- location: file:line
- issue: One sentence describing what was found
- recommendation: How to fix

If no findings: NO_FINDINGS

Perspective-Specific Guidance

视角特定指南

Perspective	Agent Focus
Completeness	Scan for markers, check checklists, verify all sections populated
Consistency	Cross-reference terms, verify links, detect contradictions
Alignment	Compare docs to code, verify implementations exist, flag hallucinations
Coverage	Map requirements to specs, check interface completeness, find gaps
Drift	Compare spec requirements to implementation, categorize drift types
Constitution	Parse rules, apply patterns/checks, report violations by level

视角	Agent聚焦点
完整性	扫描标记、检查清单、验证所有章节已填充
一致性	交叉引用术语验证链接、检测矛盾
对齐性	对比文档与代码、验证实现存在、标记虚构内容
覆盖度	将需求映射到规范、检查接口完整性、发现缺口
漂移	对比规范需求与实现、分类漂移类型
章程	解析规则、应用模式/检查、按级别报告违规

Phase 1: Parse Input and Gather Context

阶段1：解析输入并收集上下文

Analyze
```
$ARGUMENTS
```
to select validation mode (see Decision: Validation Mode)
Gather context based on mode:
- Spec Validation: Check which documents exist (PRD, SDD, PLAN), read spec files, identify cross-references
- Drift Detection: Load spec documents, identify implementation files, extract requirements and interfaces
- Constitution Validation: Check for CONSTITUTION.md at project root, parse rules by category, identify applicable scopes
- File Validation: Read target file, identify related specs or tests
- Comparison: Read both sources
Determine applicable perspectives
Present execution mode selection (see Decision: Execution Mode Selection)

分析
```
$ARGUMENTS
```
选择验证模式（参考决策：验证模式）
根据模式收集上下文：
- 规范验证：检查存在哪些文档（PRD、SDD、PLAN）、读取规范文件、识别交叉引用
- 漂移检测：加载规范文档、识别实现文件、提取需求和接口
- 章程验证：检查项目根目录下的CONSTITUTION.md、按类别解析规则、识别适用范围
- 文件验证：读取目标文件、识别相关规范或测试
- 对比验证：读取两个来源
确定适用的验证视角
呈现执行模式选择（参考决策：执行模式选择）

Standard Workflow

标准工作流

Launch ALL applicable perspectives in parallel (single response with multiple Task calls). Use the Task Delegation Template above. Continue to Synthesis.

并行启动所有适用的视角（单个响应包含多个Task调用）。使用上述任务委托模板。进入综合阶段。

Team Mode Workflow

团队模式工作流

Requires
CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS
enabled in settings.

需要在设置中启用
CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS
。

Setup

设置

Create team — derive name from target (e.g.,

validate-005

validate-drift-003

validate-constitution

)

Create one task per applicable perspective — all independent, no dependencies. Each task describes perspective focus, target files, spec context, and expected output format (Finding schema)
Spawn one validator per perspective:

Teammate	Perspective	subagent_type
`completeness-validator`	Completeness	`general-purpose`
`consistency-validator`	Consistency	`general-purpose`
`alignment-validator`	Alignment	`general-purpose`
`coverage-validator`	Coverage	`general-purpose`
`drift-validator`	Drift	`general-purpose`
`constitution-validator`	Constitution	`general-purpose`

Assign each task to its corresponding validator

Validator prompt should include: target files, spec files, project standards, expected output format (Finding schema), and team protocol: check TaskList → mark in_progress/completed → send findings to lead → claim next unblocked task when done.

创建团队 —— 从目标派生名称（例如
```
validate-005
```
、
```
validate-drift-003
```
、
```
validate-constitution
```
）
为每个适用视角创建一个任务 —— 所有任务独立，无依赖。每个任务描述视角聚焦点、目标文件、规范上下文和预期输出格式（Finding schema）
为每个视角生成一个验证器：

团队成员	视角	subagent_type
`completeness-validator`	完整性	`general-purpose`
`consistency-validator`	一致性	`general-purpose`
`alignment-validator`	对齐性	`general-purpose`
`coverage-validator`	覆盖度	`general-purpose`
`drift-validator`	漂移	`general-purpose`
`constitution-validator`	章程	`general-purpose`

为每个验证器分配对应的任务

验证器提示词应包含：目标文件、规范文件、项目标准、预期输出格式（Finding schema）和团队协议：检查TaskList → 标记in_progress/completed → 将发现结果发送给负责人 → 完成后领取下一个未阻塞的任务。

Monitoring

监控

Messages arrive automatically. If blocked: provide context via DM. After 3 retries, skip that perspective and note it.

消息自动接收。如果阻塞：通过DM提供上下文。重试3次后，跳过该视角并记录。

Shutdown

关闭

After all validators report: verify via TaskList → send sequential

shutdown_request

to each → wait for approval → TeamDelete. Continue to Synthesis.

所有验证器报告后：通过TaskList验证 → 向每个验证器发送顺序

shutdown_request

→ 等待确认 → TeamDelete。进入综合阶段。

Synthesis and Report

综合与报告

Algorithm: Deduplication

算法：去重

Applied after collecting findings from all agents/validators:

Collect all findings from all perspectives
Group by location (file:line range overlap — within 5 lines = potential overlap)
For overlapping findings: keep highest severity, merge complementary details, credit both perspectives
Sort by severity (FAIL > WARN > PASS)
Assign IDs:
```
F[N]
```
for failures,
```
W[N]
```
for warnings

收集所有Agent/验证器的发现结果后应用：

收集所有视角的发现结果
按位置分组（file:line范围重叠——5行以内视为潜在重叠）
对于重叠发现：保留最高严重程度、合并互补细节、同时标注两个视角
按严重程度排序（FAIL > WARN > PASS）
分配ID：失败项为
```
F[N]
```
，警告项为
```
W[N]
```

Report Format

报告格式

Present per Output Schema:

markdown

undefined

按照输出Schema呈现：

markdown

undefined

Validation: [target]

验证：[target]

Summary

摘要

Perspective	Pass	Warn	Fail
Completeness	X	X	X
Consistency	X	X	X
Alignment	X	X	X
Coverage	X	X	X
Drift	X	X	X
Constitution	X	X	X
Total	X	X	X

🔴 Failures (Must Fix)

ID	Finding	Recommendation
F1	Brief title (file:line)	Fix recommendation (issue description)

🟡 Warnings (Should Fix)

ID	Finding	Recommendation
W1	Brief title (file:line)	Fix recommendation (issue description)

✅ Passes

Perspective	Verified
Completeness	All sections populated, no TODO markers

视角	通过	警告	失败
完整性	X	X	X
一致性	X	X	X
对齐性	X	X	X
覆盖度	X	X	X
漂移	X	X	X
章程	X	X	X
总计	X	X	X

🔴 失败项（必须修复）

ID	发现结果	修复建议
F1	简短标题 (file:line)	修复建议 (问题描述)

🟡 警告项（建议修复）

ID	发现结果	修复建议
W1	简短标题 (file:line)	修复建议 (问题描述)

✅ 通过项

视角	已验证内容
完整性	所有章节已填充，无TODO标记

Verdict

结论

[What was validated and key conclusions]

undefined

[已验证内容和关键结论]

undefined

Mode-Specific Synthesis

模式特定综合

Drift Detection:

Categorize by drift type: Scope Creep, Missing, Contradicts, Extra
Symbols: ✅ Aligned, ❌ Missing, ⚠️ Contradicts, 🔶 Extra
Log decisions to spec README.md

Constitution Validation:

Separate by level: L1 (autofix required), L2 (manual fix required), L3 (advisory only)
L1/L2 are blocking; L3 is informational
Pattern rules: regex match. Check rules: semantic analysis

Ambiguity Detection (Spec Validation):

Detect vague patterns: hedge words ("should", "might"), vague quantifiers ("fast", "many"), open-ended lists ("etc."), undefined terms ("the system")
Score: 0-5% Excellent, 5-15% Acceptable, 15-25% Recommend clarification, 25%+ High ambiguity

漂移检测：

按漂移类型分类：范围蔓延、缺失、矛盾、额外
符号：✅ 一致、❌ 缺失、⚠️ 矛盾、🔶 额外
将决策记录到规范README.md

章程验证：

按级别区分：L1（必须自动修复）、L2（必须手动修复）、L3（仅建议）
L1/L2为阻塞项；L3为信息性内容
模式规则：正则匹配。检查规则：语义分析

模糊性检测（规范验证）：

检测模糊模式：模糊词（"should"、"might"）、模糊量词（"fast"、"many"）、开放式列表（"etc."）、未定义术语（"the system"）
评分：0-5% 优秀，5-15% 可接受，15-25% 建议澄清，25%+ 高度模糊

Integration Points

集成点

Called by
```
/start:implement
```
at phase checkpoints (drift) and completion (comparison)
Called by
```
/start:specify
```
during SDD phase for architecture alignment

在阶段检查点（漂移检测）和完成时（对比验证）被
```
/start:implement
```
调用
在SDD阶段被
```
/start:specify
```
调用以进行架构对齐

Entry Point

入口点

Parse
```
$ARGUMENTS
```
and determine validation mode (Decision: Validation Mode)
Read project context (Vision)
Gather context for the determined mode (Phase 1)
Present execution mode selection (Decision: Execution Mode Selection)
Launch validation per selected workflow (Standard or Team)
Synthesize and deduplicate findings (Synthesis)
Present report per Output Schema
Offer next steps based on findings (Decision: Next Steps)

解析
```
$ARGUMENTS
```
并确定验证模式（决策：验证模式）
读取项目上下文（愿景）
为确定的模式收集上下文（阶段1）
呈现执行模式选择（决策：执行模式选择）
根据选择的工作流启动验证（标准或团队）
综合并去重发现结果（综合阶段）
按照输出Schema呈现报告
根据发现结果提供下一步操作（决策：下一步操作）