Test Strategy

Goal

Define an appropriate test strategy for changed code, write and run tests to verify quality. Also verify the quality of the tests themselves to prevent "tests that pass but are meaningless".

Instructions

Step 1: Determine Change Type

Analyze changed files to determine the test mode.

Step 2: Identify Test Targets

• •Happy Path: Core success scenarios
• •Edge Cases: Boundary values, empty input, max values, null/undefined
• •Error Cases: Exception throwing, timeouts, invalid input
• •Integration Points: External dependency integration (mock if needed)

Step 3: Write Tests

Mode A (New Feature):

• •Follow the project's existing test patterns/framework
• •File placement: according to project conventions

Mode B (Modification):

1. •Load existing tests
2. •Gap Analysis: Do existing tests cover the new logic?
   • •No → Add new test cases
   • •API signature change → Update existing test call sites
3. •Do not delete existing tests (unless functionality is explicitly deprecated)

Mode C (Refactoring):

• •No new tests needed; run existing tests to confirm no regression

Step 4: Run Tests & Analyze

Execution:

• •Run newly written/modified tests
• •Also run related existing test suites (regression testing)

Result analysis:

Step 5: Test Quality Gate

Even if all tests pass, ⚠️ warn if the following criteria are not met.

Assertion quality:

• • Empty Test: Are there tests that only execute without any assertions?
• • Tautology: Are there always-true assertions (assertTrue(true))?
• • Insufficient assertions: Is the state sufficiently verified after execution? (minimum 1 meaningful assertion)

Mock quality:

• • Excessive mocking: Is the test target itself being mocked?
• • Unverified mocks: Are mocks set up but never verified?
• • Prefer real behavior: Even if mockable in unit tests, is there integration testing with real behavior?

Coverage gaps:

• • Missing error paths: Only happy path tested, failure paths ignored?
• • Missing boundary values: Are there tests for 0, null, empty string, max values, etc.?
• • Missing state transitions: Are all transition paths tested for stateful objects?

Step 6: Report Results

🧪 Test Report

Mode: [A: New / B: Update / C: Verify]

New tests: N added
Modified tests: M
Run result: ✅ All pass / ❌ N failed

Test Quality:
  Assertion quality: ✅ / ⚠️ (N issue(s))
  Mock quality:      ✅ / ⚠️ (N issue(s))
  Coverage gaps:     ✅ / ⚠️ (N gap(s))

[On failure]
❌ Failure list:
  - [New/Regression] Test name: Failure cause summary

[On quality warnings]
⚠️ Test Quality Warnings:
  - [Empty Test] TestUserCreate: No assertion → Add field validation for created User
  - [Coverage Gap] Error path untested: No exception verification for non-existent userId lookup

Constraints

• •Follow the project's existing test framework and patterns
• •Do not delete existing tests (unless explicitly deprecated)
• •Test names must clearly describe the behavior being tested
• •Use mocks minimally; prefer real behavior testing
• •Test quality warnings are improvement suggestions, not forced fixes — but fix if found repeatedly