AI Vibe 06 — Testing & Code Review

AI excels at writing unit tests for pure functions. Give it the function, tell it which cases to cover, and specify your test framework.

# Prompt template:
"Write Vitest unit tests for the following function.
[paste function]

Test cases to cover:
- Normal case (happy path)
- Empty input / null / undefined
- Boundary values (0, negative, very large)
- Error cases (invalid types)
- [any domain-specific edge cases]

Use describe/it blocks. Do not mock anything unless necessary."

# Example result for formatCurrency():
describe('formatCurrency', () => {
  it('formats a positive USD amount', () => {
    expect(formatCurrency(1234.56, 'USD')).toBe('$1,234.56')
  })
  it('defaults to USD when currency omitted', () => {
    expect(formatCurrency(100)).toBe('$100.00')
  })
  it('returns N/A for NaN', () => {
    expect(formatCurrency(NaN, 'USD')).toBe('N/A')
  })
  it('handles zero', () => {
    expect(formatCurrency(0, 'USD')).toBe('$0.00')
  })
  it('handles negative amounts', () => {
    expect(formatCurrency(-50, 'USD')).toBe('-$50.00')
  })
})

# Code review prompt:
"Review the following code for:
1. Correctness bugs (logic errors, off-by-one, null issues)
2. Security vulnerabilities (injection, auth bypasses, data exposure)
3. Code quality (readability, naming, duplication)
4. Performance issues (unnecessary loops, missing indexes, N+1)
5. Missing error handling

Be specific: for each issue, state the line/section,
the problem, and how to fix it.

[paste code]"

# Example targeted review:
"Review this Laravel controller method for security issues only.
 Pay attention to: authorization checks, input validation,
 SQL injection risks, and sensitive data in responses.
 [paste method]"

# When you encounter unfamiliar code:
"Explain what this code does step by step.
 Assume I understand JavaScript but am not familiar with
 this specific pattern.
 [paste code]"

# For a complex regex:
"Explain what this regex matches:
 /^(?=.*[A-Z])(?=.*\d)(?=.*[@$!%*?&])[A-Za-z\d@$!%*?&]{8,}$/
 Break down each part. Give 3 examples that match and 3 that don't."

# For a SQL query:
"Explain this SQL query in plain English.
 Describe what each JOIN does and what the final result set contains.
 [paste query]"

# For an algorithm:
"Explain this sorting algorithm. What is its time complexity?
 Walk through it with the input [3, 1, 4, 1, 5] step by step."

# JSDoc generation:
"Add JSDoc comments to all exported functions in this file.
 Include: @param (name, type, description), @returns, @throws,
          @example with a realistic usage example.
 [paste file]"

# README generation:
"Write a README section for the useCart composable below.
 Include: what it does, how to use it, all exported properties
 and methods with their types, and a complete usage example.
 [paste composable]"

# API docs:
"Document this Laravel controller as an API reference.
 For each method include: HTTP method, endpoint URL,
 request parameters, response format, and error codes."

# Security-focused review:
"Check this code for the following vulnerabilities:
- SQL injection: any raw queries with user input?
- XSS: any user content rendered as raw HTML?
- IDOR: are resource IDs validated against the authenticated user?
- Missing auth: any endpoints missing authentication middleware?
- Sensitive data: are passwords, tokens, or PII logged or returned?
- Mass assignment: are there unguarded fillable fields?

[paste Laravel controller or Vue component]"

# Input validation check:
"Is this form validation sufficient? What inputs could an attacker
 send to bypass validation or cause unexpected behavior?
 [paste validation rules]"

# E2E test prompt:
"Write a Playwright test for the user login flow:
1. Navigate to /login
2. Fill in email: test@example.com, password: secret123
3. Click the Submit button
4. Assert URL changes to /dashboard
5. Assert the page contains 'Welcome back'
6. Also test the failure case: wrong password shows error message."

# When a test fails, paste the error to AI:
"This Vitest test is failing with:
AssertionError: expected 'USD 1,234.56' to equal '$1,234.56'

Here is the test:
[paste test]

Here is the function being tested:
[paste function]

What is causing the mismatch and how do I fix the function?"

# When AI-written tests themselves are wrong:
"The test AI generated expects the wrong value.
 The function returns an object but the test checks a string.
 Here's the actual function output: { amount: 100, currency: 'USD' }
 Fix the test assertions to match the actual return shape."

AI is not a substitute for human review. It consistently misses or struggles with:

• Business logic correctness: AI doesn't know your domain rules (e.g., "a discount can't exceed the product's cost price")
• Historical context: Why a "hacky" piece of code exists and why it must stay that way
• Architectural decisions: Whether a solution fits your long-term system design
• Performance at scale: A query that's fine at 1,000 rows but breaks at 1,000,000
• Race conditions: Concurrency bugs are subtle and AI often misses them
• Third-party system quirks: Undocumented API behaviors, payment gateway edge cases
• Compliance requirements: GDPR, HIPAA, PCI-DSS specifics for your use case