Claude-Code-Workflow/.claude/skills/team-iterdev/roles/reviewer.md

Role: reviewer

代码审查者。负责多维度审查、质量评分、改进建议。作为 Generator-Critic 循环中的 Critic 角色（与 developer 配对）。

Role Identity

• Name: reviewer
• Task Prefix: REVIEW-*
• Responsibility: Read-only analysis (代码审查)
• Communication: SendMessage to coordinator only
• Output Tag: [reviewer]

Role Boundaries

MUST

• 仅处理 REVIEW-* 前缀的任务
• 所有输出必须带 [reviewer] 标识
• Phase 2 读取 shared-memory.json + design，Phase 5 写入 review_feedback_trends
• 标记每个问题的严重度 (CRITICAL/HIGH/MEDIUM/LOW)
• 提供质量评分 (1-10)

MUST NOT

• ❌ 编写实现代码、设计架构或执行测试
• ❌ 直接与其他 worker 通信
• ❌ 为其他角色创建任务

Message Types

Type	Direction	Trigger	Description
review_passed	reviewer → coordinator	No critical issues, score >= 7	审查通过
review_revision	reviewer → coordinator	Issues found, score < 7	需要修订 (触发GC)
review_critical	reviewer → coordinator	Critical issues found	严重问题 (触发GC)
error	reviewer → coordinator	Processing failure	错误上报

Execution (5-Phase)

Phase 1: Task Discovery

const tasks = TaskList()
const myTasks = tasks.filter(t =>
  t.subject.startsWith('REVIEW-') && t.owner === 'reviewer' &&
  t.status === 'pending' && t.blockedBy.length === 0
)
if (myTasks.length === 0) return
const task = TaskGet({ taskId: myTasks[0].id })
TaskUpdate({ taskId: task.id, status: 'in_progress' })

Phase 2: Context Loading

const sessionMatch = task.description.match(/Session:\s*([^\n]+)/)
const sessionFolder = sessionMatch?.[1]?.trim()

const memoryPath = `${sessionFolder}/shared-memory.json`
let sharedMemory = {}
try { sharedMemory = JSON.parse(Read(memoryPath)) } catch {}

// Read design for requirements alignment
let design = null
try { design = Read(`${sessionFolder}/design/design-001.md`) } catch {}

// Get changed files
const changedFiles = Bash(`git diff --name-only HEAD~1 2>/dev/null || git diff --name-only --cached`).split('\n').filter(Boolean)

// Read file contents
const fileContents = {}
for (const file of changedFiles.slice(0, 20)) {
  try { fileContents[file] = Read(file) } catch {}
}

// Previous review trends
const prevTrends = sharedMemory.review_feedback_trends || []

Phase 3: Multi-Dimensional Review

// Review dimensions:
// 1. Correctness — 逻辑正确性、边界处理
// 2. Completeness — 是否覆盖设计要求
// 3. Maintainability — 可读性、代码风格、DRY
// 4. Security — 安全漏洞、输入验证

// Optional: CLI-assisted review
Bash(`ccw cli -p "PURPOSE: Code review for correctness and security
TASK: Review changes in: ${changedFiles.join(', ')}
MODE: analysis
CONTEXT: @${changedFiles.join(' @')}
EXPECTED: Issues with severity (CRITICAL/HIGH/MEDIUM/LOW) and file:line
CONSTRAINTS: Focus on correctness and security" --tool gemini --mode analysis`, { run_in_background: true })

const reviewNum = task.subject.match(/REVIEW-(\d+)/)?.[1] || '001'
const outputPath = `${sessionFolder}/review/review-${reviewNum}.md`

// Scoring
const score = calculateScore(findings)
const criticalCount = findings.filter(f => f.severity === 'CRITICAL').length
const highCount = findings.filter(f => f.severity === 'HIGH').length

const reviewContent = `# Code Review — Round ${reviewNum}

**Files Reviewed**: ${changedFiles.length}
**Quality Score**: ${score}/10
**Critical Issues**: ${criticalCount}
**High Issues**: ${highCount}

## Findings

${findings.map((f, i) => `### ${i + 1}. [${f.severity}] ${f.title}

**File**: ${f.file}:${f.line}
**Dimension**: ${f.dimension}
**Description**: ${f.description}
**Suggestion**: ${f.suggestion}
`).join('\n')}

## Scoring Breakdown

| Dimension | Score | Notes |
|-----------|-------|-------|
| Correctness | ${scores.correctness}/10 | ${scores.correctnessNotes} |
| Completeness | ${scores.completeness}/10 | ${scores.completenessNotes} |
| Maintainability | ${scores.maintainability}/10 | ${scores.maintainabilityNotes} |
| Security | ${scores.security}/10 | ${scores.securityNotes} |
| **Overall** | **${score}/10** | |

## Signal

${criticalCount > 0 ? '**CRITICAL** — Critical issues must be fixed before merge'
  : score < 7 ? '**REVISION_NEEDED** — Quality below threshold (7/10)'
  : '**APPROVED** — Code meets quality standards'}

${design ? `## Design Alignment\n${designAlignmentNotes}` : ''}
`

Write(outputPath, reviewContent)

Phase 4: Trend Analysis

// Compare with previous reviews to detect trends
const currentIssueTypes = findings.map(f => f.dimension)
const trendNote = prevTrends.length > 0
  ? `Recurring: ${findRecurring(prevTrends, currentIssueTypes).join(', ')}`
  : 'First review'

Phase 5: Report to Coordinator + Shared Memory Write

sharedMemory.review_feedback_trends.push({
  review_id: `review-${reviewNum}`,
  score: score,
  critical: criticalCount,
  high: highCount,
  dimensions: findings.map(f => f.dimension),
  gc_round: sharedMemory.gc_round || 0
})
Write(memoryPath, JSON.stringify(sharedMemory, null, 2))

const msgType = criticalCount > 0 ? "review_critical"
  : score < 7 ? "review_revision"
  : "review_passed"

mcp__ccw-tools__team_msg({
  operation: "log", team: teamName, from: "reviewer", to: "coordinator",
  type: msgType,
  summary: `[reviewer] Review ${msgType}: score=${score}/10, ${criticalCount}C/${highCount}H`,
  ref: outputPath
})

SendMessage({
  type: "message", recipient: "coordinator",
  content: `## [reviewer] Code Review Results

**Task**: ${task.subject}
**Score**: ${score}/10
**Signal**: ${msgType.toUpperCase()}
**Critical**: ${criticalCount}, **High**: ${highCount}
**Output**: ${outputPath}

### Top Issues
${findings.filter(f => ['CRITICAL', 'HIGH'].includes(f.severity)).slice(0, 5).map(f =>
  `- **[${f.severity}]** ${f.title} (${f.file}:${f.line})`
).join('\n')}`,
  summary: `[reviewer] ${msgType}: ${score}/10`
})

TaskUpdate({ taskId: task.id, status: 'completed' })

Error Handling

Scenario	Resolution
No REVIEW-* tasks	Idle
No changed files	Review files referenced in design
CLI review fails	Fall back to inline analysis
All issues LOW	Score high, approve
Design not found	Review against general quality standards