---

name: audit-format description: > Design document review format definition with 100-point scoring system. Provides quality evaluation criteria, scoring items, and pass/fail logic for SOW/Spec documents. Combined with confidence markers (✓/→/?) to clarify evaluation evidence. Triggers on keywords: "SOW review", "Spec review", "document review", "design review", "90 points", "scoring", "grading".

Review Format - Design Document Scoring

Format definition for evaluating SOW/Spec document quality with 100-point scoring.

Purpose

• Objectively evaluate design document quality (sow.md, spec.md)
• Quality gate through 90-point pass/fail threshold
• Clarify evaluation evidence with confidence markers (✓/→/?)

Scoring System

Scoring Items (25 points each, 100 total)

scoring:
  accuracy:       # Accuracy score (0-25 points)
    description: "Distinction between facts and inferences, clarity of evidence"
    criteria:
      - Confidence markers (✓/→/?) are appropriately used
      - Specific evidence such as file paths and line numbers are provided
      - Reasoning is clearly stated for inferences

  completeness:   # Completeness score (0-25 points)
    description: "Coverage of required information, absence of omissions"
    criteria:
      - Acceptance criteria are specific and testable
      - Risks and mitigations are enumerated
      - Dependencies are clearly documented

  relevance:      # Relevance score (0-25 points)
    description: "Alignment with objectives, appropriateness of scope"
    criteria:
      - Goals and solutions are aligned
      - Scope is clear without excess or deficiency
      - No unnecessary features included (YAGNI)

  actionability:  # Actionability score (0-25 points)
    description: "Concrete enough for implementation, clarity of next actions"
    criteria:
      - Implementation steps are specific
      - Technical feasibility is verified
      - Next actions are clear

total: 100 points
pass_threshold: 90 points
# Rationale: 90 points ensures all 4 scoring items achieve at least
# "18-22" range (acceptable level), preventing weak areas from being
# masked by strong areas. Lower thresholds risk proceeding with
# incomplete or low-quality specifications.

Judgment Criteria

Score	Judgment	Action
90-100	✅ PASS	Proceed to next phase
70-89	⚠️ CONDITIONAL	Re-review after fixing issues
0-69	❌ FAIL	Major revision needed

Output Format

Review Result Template

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

## 📋 Design Document Review Result

### Target Documents
- SOW: `{sow_path}`
- Spec: `{spec_path}`

### Total Score: {total}/100 {judgment_icon}

| Item | Score | Evaluation |
|------|-------|------------|
| Accuracy | {accuracy}/25 | {✓/→/?} |
| Completeness | {completeness}/25 | {✓/→/?} |
| Relevance | {relevance}/25 | {✓/→/?} |
| Actionability | {actionability}/25 | {✓/→/?} |

### Judgment: {PASS/CONDITIONAL/FAIL}

---

### 📊 Detailed Evaluation

#### Accuracy: {score}/25
**Strengths**:
- [✓] {Good points with evidence}

**Areas for Improvement**:
- [→] {Points needing improvement with reason}

#### Completeness: {score}/25
**Strengths**:
- [✓] {Good points with evidence}

**Areas for Improvement**:
- [?] {Missing information}

#### Relevance: {score}/25
**Strengths**:
- [✓] {Good points with evidence}

**Areas for Improvement**:
- [→] {Scope issues}

#### Actionability: {score}/25
**Strengths**:
- [✓] {Good points with evidence}

**Areas for Improvement**:
- [?] {Implementation concerns}

---

### 🔗 SOW ↔ Spec Consistency Check

| Check Item | Status | Details |
|------------|--------|---------|
| Acceptance Criteria → Functional Requirements | {✓/→/?} | {details} |
| Risks → Mitigations | {✓/→/?} | {details} |
| Dependency Alignment | {✓/→/?} | {details} |
| Test Plan Coverage | {✓/→/?} | {details} |

---

### 📝 Requested Modifications

#### 🔴 Required Fixes (Pass Condition)
1. {Specific modification content}
2. {Specific modification content}

#### 🟡 Recommended Fixes
1. {Improvement suggestion}
2. {Improvement suggestion}

---

### Next Action

{If PASS}
✅ Can proceed to implementation phase (/code)

{If CONDITIONAL}
⚠️ Please re-review after addressing the requested modifications above

{If FAIL}
❌ Major revision of SOW/Spec is required. Please re-plan with /think

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Scoring Guidelines

Accuracy Evaluation Criteria

Score	Criteria
23-25	All claims have ✓ markers and concrete evidence
18-22	Most claims have evidence, but some have → or ?
13-17	Multiple areas with ambiguous evidence
0-12	Many claims without evidence

Completeness Evaluation Criteria

Score	Criteria
23-25	All sections are comprehensively documented
18-22	Main sections complete, minor omissions in details
13-17	Important sections have gaps
0-12	Required sections significantly lacking

Relevance Evaluation Criteria

Score	Criteria
23-25	Goals and solutions fully aligned, appropriate scope
18-22	Generally aligned, minor scope excess or deficiency
13-17	Partial misalignment, scope issues present
0-12	Large gap between goals and solutions

Actionability Evaluation Criteria

Score	Criteria
23-25	Specific enough to start implementation immediately
18-22	Most parts implementable, some require confirmation
13-17	Additional investigation needed before implementation
0-12	Difficult to implement, significant detail needed

SOW ↔ Spec Consistency Checklist

Required Check Items

1. Acceptance Criteria → Functional Requirements Mapping

   • Each acceptance criterion in SOW maps to FR-xxx in Spec
   • No mapping gaps

2. Risks → Mitigations Correspondence

   • Risks identified in SOW are addressed in Spec
   • Any newly identified risks

3. Dependency Alignment

   • SOW dependencies match Spec dependency section
   • No additional dependencies introduced

4. Test Plan Coverage

   • SOW acceptance criteria covered by Spec test scenarios
   • Test priorities are appropriate

Usage

Reference from Agents

# agents/reviewers/sow-spec.md
skills:
  - audit-format
  - readability-review
  - code-principles

Direct Invocation

Auto-triggers on keywords "SOW review", "design review", "90-point threshold", etc.

Related Documents

• [@~/.claude/agents/reviewers/sow-spec.md] - SOW/Spec specialized reviewer
• [@~/.claude/commands/think.md] - SOW/Spec generation command
• [@~/.claude/skills/readability-review/SKILL.md] - Readability evaluation criteria

---

Version History

• 1.0.0 (2025-12-08): Initial release - 100-point scoring + confidence marker integration