majiayu000 Gate 6: Testing Verification
"Tests are proof of understanding. If you can't test it, do you really understand it?"
Purpose
This gate encourages juniors to write tests for their features. Unlike the Ownership Gate, this does NOT block completion - it issues warnings to encourage the testing habit.
Gate Status
- PASS — Tests exist and cover critical paths
- WARNING — No tests or insufficient coverage (can proceed with note)
Note: This gate does NOT block. The goal is to build the testing habit through encouragement, not enforcement.
Gate Questions
Ask in sequence:
Question 1: Test Existence
"What tests did you write for this feature?"
Looking for:
- At least one test file exists
- Tests are actually running (not skipped)
- Tests are meaningful (not just
expect(true).toBe(true))
If no tests:
"I noticed there aren't tests for this feature. Testing isn't required to complete, but it's a habit worth building. What would you test if you had time?"
Question 2: Coverage Strategy
"What does your test prove about this feature?"
Looking for:
- Happy path covered
- At least one edge case considered
- Error states (if applicable)
Follow-up:
"If I broke [specific part], which test would catch it?"
Question 3: Test Quality
"Show me your most important test. What behavior does it verify?"
Looking for:
- Testing behavior, not implementation
- Clear test names
- AAA pattern (Arrange, Act, Assert)
Response Templates
If PASS (Tests Exist)
✅ TESTING GATE: PASSED
Nice work including tests! I see you covered:
- [Specific test they wrote]
- [Edge case they handled]
Key strength: [Something they did well]
Consider adding: [One suggestion for future]
Moving to code review...
If WARNING (No Tests)
⚠️ TESTING GATE: WARNING
No tests found for this feature. That's okay - we can proceed.
But here's why tests matter:
1. **Interview Gold**: "I implemented tests for critical flows..."
2. **Confidence**: Know your changes don't break things
3. **Documentation**: Tests show how code should be used
Quick win for next time:
- Test the happy path first
- Add one edge case
- That's already better than most!
Proceeding to code review...
If WARNING (Weak Tests)
⚠️ TESTING GATE: WARNING
Tests exist but could be stronger:
**Issue**: [What's missing or weak]
**Question**: "If [scenario], would your tests catch it?"
This doesn't block you, but consider:
- [Specific improvement suggestion]
Proceeding to code review...
What Makes a Good Test Suite?
| Level | Coverage | Characteristics |
|---|---|---|
| Minimal | 1-2 tests | Happy path only |
| Good | 3-5 tests | Happy path + main edge cases |
| Strong | 5-10 tests | Happy path + edge cases + error states |
| Interview-Ready | Full pyramid | Unit + Integration + E2E for critical flows |
Socratic Guidance
If they want to add tests but don't know where to start:
- "What's the ONE thing that would be really bad if it broke?"
- "What input would a user never send but a hacker might?"
- "What happens when the server is slow or returns an error?"
Stack-Specific Hints
| Stack | Suggestion |
|---|---|
| Vite + React | "Vitest + React Testing Library is fast and integrated" |
| Next.js | "Vitest or Jest works great with Next" |
| API/Backend | "Test your endpoints with supertest or native HTTP" |
| Python | "pytest is the standard - simple and powerful" |
Interview Connection
"Testing is interview gold."
When they pass this gate with tests:
- Note it for their STAR story
- "You can talk about your testing strategy in interviews"
- "What percentage coverage did you achieve?" (for resume)
When they skip tests:
- "Next time, even 2-3 tests makes a huge difference for your portfolio"
- "Employers love seeing test files in your repo"
Why WARNING Not BLOCKING?
- Encouragement > Enforcement: Build the habit through positive reinforcement
- Some features are trivial: Not everything needs tests
- Time constraints exist: Production pressure is real
- Learning curve: Testing has a learning curve; don't block progress
The goal is to make testing feel valuable, not punitive.