Files
Claude-Code-Workflow/.claude/skills/review-code/specs/quality-standards.md
catlog22 29c8bb7a66 feat: Add orchestrator and state management for code review process
- Implemented orchestrator logic to manage code review phases, including state reading, action selection, and execution loop.
- Defined state schema for review process, including metadata, context, findings, and execution tracking.
- Created action catalog detailing actions for context collection, quick scan, deep review, report generation, and completion.
- Established error recovery strategies and termination conditions for robust review handling.
- Developed issue classification and quality standards documentation to guide review severity and categorization.
- Introduced review dimensions with detailed checklists for correctness, security, performance, readability, testing, and architecture.
- Added templates for issue reporting and review reports to standardize output and improve clarity.
2026-01-13 14:39:16 +08:00

4.6 KiB

Quality Standards

代码审查质量标准。

When to Use

Phase Usage Section
action-generate-report 质量评估 Quality Dimensions
action-complete 最终评分 Quality Gates

Quality Dimensions

1. Completeness (完整性) - 25%

评估审查覆盖的完整程度

Score Criteria
100% 所有维度审查完成,所有高风险文件检查
80% 核心维度完成,主要文件检查
60% 部分维度完成
< 60% 审查不完整

检查点:

  • 6 个维度全部审查
  • 高风险区域重点检查
  • 关键文件覆盖

2. Accuracy (准确性) - 25%

评估发现问题的准确程度

Score Criteria
100% 问题定位准确,分类正确,无误报
80% 偶有分类偏差,定位准确
60% 存在误报或漏报
< 60% 准确性差

检查点:

  • 问题行号准确
  • 严重程度合理
  • 分类正确

3. Actionability (可操作性) - 25%

评估建议的实用程度

Score Criteria
100% 每个问题都有具体可执行的修复建议
80% 大部分问题有清晰建议
60% 建议较笼统
< 60% 缺乏可操作建议

检查点:

  • 提供具体修复建议
  • 包含代码示例
  • 说明修复优先级

4. Consistency (一致性) - 25%

评估审查标准的一致程度

Score Criteria
100% 相同问题相同处理,标准统一
80% 基本一致,偶有差异
60% 标准不太统一
< 60% 标准混乱

检查点:

  • ID 格式统一
  • 严重程度标准一致
  • 描述风格统一

Quality Gates

Review Quality Gate

Gate Overall Score Action
Excellent ≥ 90% 高质量审查
Good ≥ 80% 合格审查
Acceptable ≥ 70% 基本可接受
Needs Improvement < 70% 需要改进

Code Quality Gate (Based on Findings)

Gate Condition Recommendation
Block Critical > 0 禁止合并,必须修复
Warn High > 3 需要团队讨论
Caution Medium > 10 建议改进
Pass 其他 可以合并

Report Quality Checklist

Structure

  • 包含审查概览
  • 包含问题统计
  • 包含高风险区域
  • 包含问题详情
  • 包含修复建议

Content

  • 问题描述清晰
  • 文件位置准确
  • 代码片段有效
  • 修复建议具体
  • 优先级明确

Format

  • Markdown 格式正确
  • 表格对齐
  • 代码块语法正确
  • 链接有效
  • 无拼写错误

Validation Function

function validateReviewQuality(state) {
  const scores = {
    completeness: 0,
    accuracy: 0,
    actionability: 0,
    consistency: 0
  };
  
  // 1. Completeness
  const dimensionsReviewed = state.reviewed_dimensions?.length || 0;
  scores.completeness = (dimensionsReviewed / 6) * 100;
  
  // 2. Accuracy (需要人工验证或后续反馈)
  // 暂时基于有无错误来估算
  scores.accuracy = state.error_count === 0 ? 100 : Math.max(0, 100 - state.error_count * 20);
  
  // 3. Actionability
  const findings = Object.values(state.findings).flat();
  const withRecommendations = findings.filter(f => f.recommendation).length;
  scores.actionability = findings.length > 0 
    ? (withRecommendations / findings.length) * 100 
    : 100;
  
  // 4. Consistency (检查 ID 格式等)
  const validIds = findings.filter(f => /^(CORR|SEC|PERF|READ|TEST|ARCH)-\d{3}$/.test(f.id)).length;
  scores.consistency = findings.length > 0 
    ? (validIds / findings.length) * 100 
    : 100;
  
  // Overall
  const overall = (
    scores.completeness * 0.25 +
    scores.accuracy * 0.25 +
    scores.actionability * 0.25 +
    scores.consistency * 0.25
  );
  
  return {
    scores,
    overall,
    gate: overall >= 90 ? 'excellent' :
          overall >= 80 ? 'good' :
          overall >= 70 ? 'acceptable' : 'needs_improvement'
  };
}

Improvement Recommendations

If Completeness is Low

  • 增加扫描的文件范围
  • 确保所有维度都被审查
  • 重点关注高风险区域

If Accuracy is Low

  • 提高规则精度
  • 减少误报
  • 验证行号准确性

If Actionability is Low

  • 为每个问题添加修复建议
  • 提供代码示例
  • 说明修复步骤

If Consistency is Low

  • 统一 ID 格式
  • 标准化严重程度判定
  • 使用模板化描述