codeprobe-testing

name	codeprobe-testing
description	Audits code for test quality and coverage issues — missing tests, test smells, poor test structure, mock abuse, coverage gaps, and fragile test data. Identifies weaknesses in the test suite and generates fix prompts. Trigger phrases: "test quality", "test audit", "test review", "coverage check", "missing tests", "test quality audit".
allowed-tools	["Read","Grep","Glob","Bash"]

Standalone Mode

If invoked directly (not via the orchestrator), you must first:

1. Read ../codeprobe/shared-preamble.md (resolve relative to this SKILL.md's location — the sibling codeprobe skill directory — not the user's project) for the output contract, execution modes, and constraints.
2. Load applicable reference files from ../codeprobe/references/ (same resolution) based on the project's tech stack.
3. Default to full mode unless the user specifies otherwise.

Test Quality & Coverage Auditor

Domain Scope

This sub-skill detects test quality and coverage issues across six categories:

1. Missing Tests — Public methods without corresponding tests, critical business logic untested.
2. Test Smells — Tests with no assertions, testing implementation details, brittle tests.
3. Test Structure — Missing Arrange-Act-Assert separation, poor naming, testing too many things.
4. Mock Abuse — Mocking the system under test, mock returning mocks, over-mocking.
5. Coverage Gaps — No tests for error paths, authorization logic, edge cases.
6. Test Data — Hardcoded IDs, fragile fixtures, environment-dependent tests.

---

What It Does NOT Flag

• Missing tests for trivial getters/setters or pure DTOs — these add testing overhead without meaningful coverage.
• Framework-generated test stubs that are empty but clearly scaffolded (e.g., Laravel's ExampleTest.php, Create React App's App.test.js) — these are starting points, not abandoned tests.
• Integration/E2E test suites that intentionally don't follow unit test conventions — different test levels have different design constraints.
• Test execution speed — this sub-skill assesses test design quality, not runtime performance.
• Tests in vendor/, node_modules/, or other dependency directories.

---

Detection Instructions

Severity Ceiling

No finding from this sub-skill should ever be classified as Critical. Missing tests, even for critical business logic, are a maintainability and risk issue (Major), not a confirmed production defect. The highest severity this sub-skill may assign is Major. Follow the severity column in each detection table exactly — do not escalate beyond it.

Missing Tests

ID Prefix	What to Detect	How to Detect	Severity
TEST	Public methods with no corresponding test	Scan source directories for public class methods. For each, check if a corresponding test file/method exists. Use test file naming conventions: test_, _test., .test., Test[A-Z], spec_, _spec., .spec. (matching file_stats.py patterns).	Major
TEST	Critical business logic untested	Identify classes/methods handling payments, authentication, authorization, order processing, or data mutations. Check whether these have dedicated test coverage.	Major
TEST	Edge cases unaddressed	When tests exist for a method, check whether they cover: null/empty inputs, boundary values (0, -1, max), error cases, and the happy path. Flag methods with only happy-path tests.	Minor

Test Smells

ID Prefix	What to Detect	How to Detect	Severity
TEST	Tests with no assertions	Search test methods for assertion calls (assert, expect, should, verify). Flag test methods that execute code but never assert on the outcome — they only verify "no exception thrown."	Major
TEST	Tests testing implementation details	Tests that mock every dependency and only verify call order/counts rather than outcomes. Tests that break when internal implementation changes but behavior stays the same. Look for excessive ->expects()->method()->with() chains or toHaveBeenCalledWith without checking return values.	Minor
TEST	Brittle tests coupled to external state	Tests that depend on database state not set up in the test, file system paths, network calls, or system time without mocking. Look for raw SQL in tests, file_exists() checks, HTTP calls without mocking.	Minor
TEST	Tests dependent on execution order	Tests that pass individually but fail when run together (or vice versa). Look for shared mutable state between test methods: class-level properties modified in tests, database records not cleaned up.	Major

Test Structure

ID Prefix	What to Detect	How to Detect	Severity
TEST	Missing Arrange-Act-Assert separation	Test methods where setup, execution, and assertion are interleaved rather than clearly separated. Multiple act+assert cycles in one test.	Minor
TEST	Test names that don't describe the scenario	Test methods named test1, testFunction, it_works, or using generic names that don't describe the input condition and expected outcome. Good: test_empty_cart_returns_zero_total. Bad: testCalculate.	Minor
TEST	Single test testing too many things	Test methods with 5+ assertions on unrelated outcomes, or that test multiple scenarios in sequence. Should be split into focused tests.	Minor

Mock Abuse

ID Prefix	What to Detect	How to Detect	Severity
TEST	Mocking the system under test	Test creates a mock/partial mock of the class being tested. The test is testing the mock, not the actual code. Look for $this->createPartialMock(ClassName::class) or jest.spyOn(sut, 'method') where sut is the class being tested.	Major
TEST	Mock returning mocks	Mock objects configured to return other mock objects, creating deep mock chains. $mock->method('getUser')->willReturn($userMock) where $userMock->method('getProfile')->willReturn($profileMock).	Major
TEST	Over-mocking making tests pass regardless	Tests where every dependency is mocked and the mocks return exactly what the code expects, making the test a tautology. If you change the implementation logic, the test still passes because the mocks drive the result.	Minor

Coverage Gaps

ID Prefix	What to Detect	How to Detect	Severity
TEST	No tests for error/exception paths	Methods with try/catch blocks or error handling where tests only cover the success path. No test triggers the catch/error branch.	Minor
TEST	No tests for authorization logic	Permission checks, policy methods, gate definitions, middleware authorization — code that controls access but has no dedicated tests.	Major
TEST	No edge case tests	Functions handling arrays/collections without tests for empty input. Numeric functions without tests for zero, negative, or boundary values. String functions without tests for empty string, unicode, or very long input.	Minor

Test Data

ID Prefix	What to Detect	How to Detect	Severity
TEST	Hardcoded IDs that may collide	Tests using hardcoded numeric IDs ($userId = 42, id: 1) instead of factory-generated values. These collide in parallel test runs or with seeded data.	Minor
TEST	Fragile factory/fixture setup	Tests with complex inline data setup that duplicates across multiple tests instead of using factories/fixtures/builders.	Minor
TEST	Tests relying on specific database state	Tests that assume certain records exist in the database without creating them in the test setup. Depends on seeders or previous test execution.	Minor

---

ID Prefix & Fix Prompt Examples

All findings use the TEST- prefix, numbered sequentially: TEST-001, TEST-002, etc.

Fix Prompt Examples

• "Write a test for OrderService@calculateTotal that covers: empty cart (expect 0), single item, multiple items, and item with discount. Use OrderFactory for test data. Place in tests/Unit/Services/OrderServiceTest.php."
• "The test test_user_can_login at tests/Feature/AuthTest.php:25 has no assertions — it only calls the login endpoint. Add assertStatus(200), assertAuthenticated(), and assertJsonStructure(['token']) assertions."
• "In tests/Unit/PaymentServiceTest.php:40, the mock chain is mocking too deeply. Create a concrete FakePaymentGateway that implements the gateway interface and returns predictable responses instead of nested mock returns."
• "Replace the hardcoded user ID 42 in tests/Feature/OrderTest.php:15 with User::factory()->create()->id to prevent test collisions in parallel test runs."