Learn why to review implementation even with 100% test coverage

Why Review Implementation Even with 100% Test Coverage?

Overview

This learning module helps junior developers understand why, as of today in 2026, human code review remains essential even when AI-generated code has 100% unit test coverage. Tests only verify code doesn't crash - humans must verify correctness, readability, and maintainability.

Instructions

Present the following content interactively. After each section, use AskUserQuestion to offer 3 options:

• "Continue to next section"
• "Show me an example"
• "I have a question"

---

Section 1: What 100% Test Coverage Actually Proves (and Doesn't)

Explain what test coverage measures and its limitations:

What 100% test coverage proves:

• Every line of code is executed during tests
• The code doesn't crash during execution

That's it. Coverage doesn't verify correctness - test assertions could be inadequate, wrong, or missing entirely.

What 100% test coverage does NOT prove:

• The code behaves correctly (assertions could be wrong)
• The code meets functional requirements (assertions could be inadequate)
• The code is easy to understand and intuitive
• The code will be easy to modify later
• The variable and function names are clear
• The solution is appropriately simple

The key insight:

Tests answer "Does it crash?" not necessarily "Does it work?" or "Is it good code?" All three questions matter.

---

Section 2: Why Test Cases Need Review

Since 100% coverage only proves code doesn't crash, the test assertions themselves are critical - and they need human review.

The Problem with Trusting Tests Blindly

// This test "passes" but proves nothing
test('should process order', () => {
  const result = processOrder(mockOrder);
  expect(result).toBeDefined();  // Weak assertion - just checks it exists
});

// This test passes but has wrong expectation
test('should calculate total', () => {
  const total = calculateTotal([10, 20, 30]);
  expect(total).toBe(50);  // Wrong! Should be 60
});

// This test is missing assertions entirely
test('should update user', async () => {
  await updateUser({ id: 1, name: 'John' });
  // No assertions - test passes if it doesn't crash
});