Write Tests
Converts test scenarios into actual tests. Runs after /plan-tests (creates scenarios) and /build-feature (implements code).
Inputs
- •Test scenarios doc from test scenarios location
- •Implemented feature location (service, controller, module)
Workflow
- •Read test scenarios doc
- •Read implemented code (service methods, controller endpoints, entities)
- •Detect gaps (see below)
- •Write test file following patterns in
references/test-patterns.template.md - •Run tests
- •Report: pass/fail count, any gaps found
Gap Detection
Compare scenarios against implementation. Flag:
| Gap Type | Example |
|---|---|
| Missing implementation | Scenario says "should retry 3x on failure" but no retry logic exists |
| Missing test coverage | Code has error handling not covered by any scenario |
| Behavior mismatch | Scenario expects order marked failed when inventory insufficient, but implementation marks it pending |
Report gaps before writing tests. Ask: proceed anyway, or update scenarios/implementation first?
Mocking Rules
Mock only external APIs:
typescript
jest.spyOn(ExternalClient.prototype, 'method').mockResolvedValue('...');
Never mock: internal services, repositories, domain logic.
Factory Usage
Use factories with overrides:
typescript
const item = await repository.save(itemFactory({ isActive: true }));
Naming Conventions
Describe blocks: Feature name or endpoint
typescript
describe('Order Creation', () => {
describe('POST /v1/orders', () => {
Nested describe: Context from scenario
typescript
describe('when valid data received', () => {
describe('when duplicate request received', () => {
Test cases: Match scenario descriptions
typescript
test('should create order when valid data received', async () => {
test('should return existing order without creating duplicate', async () => {
Test Consolidation
Before writing, check existing test files for overlap. Merge when:
- •Tests share setup AND action
- •One test is subset of another
- •Multiple aspects can verify in single test
Output
- •Test file(s) created/updated
- •Test run results
- •Gap report (if any found)
Final Step: Code Review
After tests pass, spawn the code-reviewer agent to review the implementation + test code. It will check:
- •Bugs and logic errors
- •Pattern violations (against CLAUDE.md and coding-patterns)
- •Code quality issues
Address critical/important issues before considering the feature ready for PR.