Skip to content

Task Review

Analysis and decision-making workflows separate from implementation.

Overview

The /task-review command provides a dedicated workflow for analysis tasks, separate from the implementation-focused /task-work command.

When to Use

Use /task-review for Use /task-work for
"Should we implement X?" "Implement feature X"
"Review architecture of X" "Fix bug in X"
"Assess technical debt in X" "Add tests for X"
"Security audit of X" "Refactor X"

Review Modes

The /task-review command supports five specialized modes:

1. Architectural Review

Purpose: Evaluate system design against SOLID/DRY/YAGNI principles

Output:

  • Architecture Assessment (scored 0-100)
  • SOLID Compliance (0-10 per principle)
  • DRY Adherence (0-10)
  • YAGNI Compliance (0-10)
  • Design Patterns Analysis
  • Technical Debt Inventory
  • Recommendations (Keep/Refactor/Rewrite)

2. Code Quality Review

Purpose: Assess code maintainability, complexity, test coverage

Output:

  • Code Metrics (complexity, LOC, duplication, coverage)
  • Quality Issues (code smells, anti-patterns)
  • Maintainability Score (0-10)
  • Refactoring Recommendations

3. Decision Analysis

Purpose: Evaluate options and provide decision recommendation

Output:

  • Current Situation Assessment
  • Root Cause Analysis (if applicable)
  • Option Evaluation Matrix
  • Recommended Decision with Rationale

4. Technical Debt Assessment

Purpose: Inventory and prioritize technical debt

Output:

  • Technical Debt Inventory
  • Priority Matrix (effort vs impact)
  • Remediation Roadmap
  • Quick Wins vs Strategic Improvements

5. Security Audit

Purpose: Security vulnerability assessment

Output:

  • Security Findings (OWASP Top 10 mapping)
  • Vulnerability Severity Ratings
  • Remediation Recommendations
  • Compliance Assessment (if applicable)

Depth Levels

Depth Duration Use For
quick 15-30 min Initial assessment, sanity checks
standard 1-2 hours Regular reviews, architecture assessments
comprehensive 4-6 hours Security audits, critical decisions

Usage Examples

Example 1: Architectural Review

# Create review task
/task-create "Review authentication architecture" task_type:review

# Execute architectural review
/task-review TASK-002 --mode=architectural --depth=standard

# Decision checkpoint (automated)
# [A]ccept - Approve findings, move to IN_REVIEW
# [R]evise - Request deeper analysis
# [I]mplement - Create implementation task based on recommendations
# [C]ancel - Discard review

Example 2: Security Audit

# Create security audit task
/task-create "Security audit of payment processing" task_type:review

# Execute comprehensive security review
/task-review TASK-SEC-D7E2 --mode=security --depth:comprehensive

# Review identifies 12 vulnerabilities
# Choose [I]mplement to create implementation task

# Implement fixes
/task-work TASK-IMP-E8F3

# Verification review
/task-create "Verify security fixes" task_type:review
/task-review TASK-VER-F9G4 --mode=security --depth=standard

Model Selection

The command automatically selects the optimal Claude model based on review mode and depth:

When Opus 4.5 Is Used

  • Security reviews (all depths) - Security breaches cost $100K-$10M
  • Decision analysis (standard/comprehensive) - Complex trade-offs
  • Comprehensive architectural reviews - Thorough analysis
  • Comprehensive technical debt - Nuanced prioritization

Cost: $0.45-$1.65 per review (67% premium vs Sonnet)

When Sonnet 4.5 Is Used

  • Quick reviews (except security) - Speed matters
  • Code quality reviews (all depths) - Metrics are objective
  • Standard architectural reviews - Pattern-based analysis
  • Standard technical debt - Straightforward prioritization

Cost: $0.09-$0.68 per review

Integration with /task-work

Review tasks seamlessly integrate with implementation tasks:

Complete Workflow:

  1. Review → Identify what needs to be done
  2. [I]mplement → Creates implementation task automatically
  3. Implement → Execute via /task-work
  4. Verify → Optional verification review

Task Review Workflow Guide - Complete documentation.

Automatic Review Task Detection

When creating tasks with /task-create, the system automatically detects review/analysis tasks and suggests using /task-review.

Detection Criteria (any of the following):

  1. task_type:review parameter
  2. decision_required:true parameter
  3. Review-related tags: architecture-review, code-review, decision-point, assessment
  4. Title keywords: review, analyze, evaluate, assess, audit, investigation

Automatic Review Task Detection - Full documentation.

Next Steps