Code Review Assistant

Code Review Assistant

Perform systematic code reviews to ensure quality, maintainability, and adherence to project standards.

Purpose

This skill evaluates code quality across multiple dimensions: clarity, correctness, test coverage, performance, and alignment with project architecture. Use this skill at the "Verify" step of the development cycle and during phase acceptance.

When to Use This Skill

Invoke when:

• Completing a feature or function
• Before committing code
• During acceptance phase
• After refactoring
• Reviewing pull requests
• Checking code quality

Review Checklist

1. Code Quality

Clarity and Readability:

• ✅ Clear, descriptive function and variable names
• ✅ Functions are small and focused ( 30 lines):

• Extract smaller functions
• Single Responsibility Principle

Deep Nesting (> 3 levels):

• Use early returns (guard clauses)
• Extract complex logic

Code Duplication:

• Extract shared logic
• Create utility functions

Magic Numbers:

// Bad
if (user.status === 2) {
}

// Good
const STATUS_ACTIVE = 2;
if (user.status === STATUS_ACTIVE) {
}

// Better
enum UserStatus {
  PENDING = 0,
  INACTIVE = 1,
  ACTIVE = 2,
  SUSPENDED = 3,
}
if (user.status === UserStatus.ACTIVE) {
}

TypeScript Issues

Using any:

// Bad
function process(data: any) {}

// Good
function process(data: UserData) {}

Missing Return Types:

// Bad
function getUser(id) {}

// Good
function getUser(id: string): User | null {}

Testing Issues

Missing Edge Cases:

• Test negative values
• Test boundary values
• Test invalid inputs

Unclear Test Names:

// Bad
it('test1', () => {});

// Good
it('returns null when position is out of bounds', () => {});

Common Architecture Patterns

Separation of Concerns

Good:

// Service: Pure business logic
class OrderService {
  calculateTotal(items: OrderItem[]): number {
    // Pure calculation
  }
}

// Controller: HTTP handling
class OrderController {
  async handleRequest(req: Request): Promise {
    const total = this.orderService.calculateTotal(req.body.items);
    return Response.json({ total });
  }
}

Bad:

// Mixed concerns
class OrderService {
  async calculateTotal(items: OrderItem[], res: Response): Promise {
    // Calculate AND send response (tight coupling)
  }
}

Event-Driven Communication

Good:

// Emitter
this.eventBus.emit('order-created', { orderId, userId });

// Listener (in another service)
this.eventBus.on('order-created', this.sendConfirmationEmail, this);

Bad:

// Direct coupling
this.emailService.sendConfirmationEmail(); // Called from order creation

Resource Cleanup

Good:

destroy(): void {
  this.eventBus.off();
  this.timers.forEach(t => clearTimeout(t));
  this.connections.forEach(c => c.close());
}

Bad:

destroy(): void {
  // No cleanup - memory leak!
}

Approval Criteria

✅ Approved

Code meets all quality standards:

• No critical issues
• Test coverage meets targets
• Follows project conventions
• Clear and maintainable
• Well-documented

⚠️ Approve with Changes

Code is functional but has minor issues:

• Some moderate issues to address
• Coverage slightly below target
• Minor refactoring beneficial
• Documentation incomplete

Action: Fix issues before next phase

❌ Changes Required

Code has significant issues:

• Critical bugs or security issues
• Missing test coverage
• Violates architecture
• Major refactoring needed
• Unreadable or unmaintainable

Action: Rework before proceeding

Critical Reminders

• Review code, not the person
• Provide actionable, specific feedback
• Explain WHY something is an issue
• Suggest concrete improvements
• Acknowledge good practices
• Be constructive and respectful

Code review is not about perfection—it's about continuous improvement and maintaining quality standards while shipping working software.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: MichaelD-42
• Source: MichaelD-42/agentic-skills
• License: MIT

Install and usage instructions live in the source repository linked above.