create-test（create 测试）

Create Test

Overview

Generate unit tests for code components. The command adapts to backend or frontend code automatically based on the codebase structure and file patterns.

Definitions

• {component}: Component, class, function, or file to test. Can be specified as:

- Class name (e.g., OAuthService, UserProfile) - File path (e.g., auth-utils.ts, src/services/payment.ts) - Component name (e.g., LoginForm, UserDashboard)

• Test Type: Type of test to generate (unit, integration, e2e). Currently focuses on unit tests.
• Test Framework: Testing library in use (Jest, Vitest, pytest, unittest, Mocha, xUnit, NUnit, MSTest, etc.)
• Codebase Type: Classification of codebase structure:

- Backend: API routes, services, models, utilities (typically Node.js, Python, Java, etc.) - Frontend: Components, hooks, utilities (typically React, Vue, Angular, etc.)

• Test File Naming Convention: Pattern for test file names (e.g., *.test.ts, *_test.py, *.spec.js, *Test.cs, *Tests.cs)
• Arrange-Act-Assert (AAA) Pattern: Test structure: Arrange (setup), Act (execute), Assert (verify)

Prerequisites

Before proceeding, verify:

1. Component Exists: Verify the component/class/file exists in the codebase

- Use codebase_search or grep to locate the component - Use read_file to verify the file is readable - If component cannot be found, STOP and report error: "Component {component} not found in codebase." - File Operation Standards: File operations should follow best practices. These standards are documented in AGENTS.md §3 Operational Boundaries if AGENTS.md exists, but apply universally regardless.

1. Codebase Structure is Understandable: Verify codebase structure is readable and navigable

- Use list_dir to explore directory structure - Identify project root, source directories, test directories - If codebase structure is unclear, STOP and ask user for clarification on project structure.

1. Test Framework is Identifiable: Verify test framework can be detected

- Check for test framework configuration files (e.g., jest.config.js, vitest.config.ts, pytest.ini, .csproj, .sln) - Look for existing test files to identify framework and naming patterns - For.NET projects, check for .csproj files with test framework references (xUnit, NUnit, MSTest) - If test framework cannot be determined, STOP and report: "Unable to identify test framework. Please specify test framework or add configuration file."

Steps

1. Check for Spec and read Contract scenarios (if exists)

- Determine feature domain from component: - Extract feature domain from component path or name - Example: src/services/auth/OAuthService.ts → feature domain might be authentication or user-auth - If unclear, ask user for feature domain - Check for Spec: - Use glob_file_search to find specs: pattern **/specs/*/spec.md - Match component to relevant spec by feature domain - If spec found at specs/{FEATURE_DOMAIN}/spec.md, proceed to read it - If no spec found, skip to Step 2 (will use code analysis) - Read Spec Contract section: - Use read_file to read the spec - Parse Contract section - Extract Scenarios subsection - Parse Gherkin scenarios: - For each scenario block: - Extract scenario name/title - Extract Given steps (preconditions) - Extract When step (action/trigger) - Extract Then steps (expected outcomes) - Store scenarios for test generation - Error Handling: - If spec exists but has no Scenarios: Note it and fall back to code analysis - If Spec is malformed: Note it and fall back to code analysis

1. Detect codebase type

- Analyze project structure: - Use list_dir to explore root directory structure - Look for patterns indicating backend (e.g., src/routes/, src/services/, src/models/) - Look for patterns indicating frontend (e.g., src/components/, src/hooks/, src/composables/, src/pages/, .vue files, .tsx/.jsx files) - Check for package.json, requirements.txt, or other dependency files to identify language/framework - Identify if backend or frontend: - Backend indicators: API routes, database models, service layers, server configuration - Frontend indicators: UI components, hooks/composables, pages, client-side routing, .vue files (Vue), .tsx/.jsx files (React) - If both present, determine primary focus based on component location - Determine test framework in use: - Use glob_file_search to find test configuration files: **/jest.config.*, **/vitest.config.*, **/pytest.ini, **/*.csproj, **/*.sln, etc. - Use glob_file_search to find existing test files: **/*.test.*, **/*_test.*, **/*.spec.*, **/*Test.cs, **/*Tests.cs - For.NET projects, check .csproj files for test framework packages (xUnit, NUnit, MSTest) - Read configuration files if found to understand framework setup - Identify naming convention from existing test files - Check existing test patterns and conventions: - Review existing test files to understand structure, imports, and patterns - Note how tests are organized (same directory, separate test directory, etc.) - Identify common test utilities or helpers - Error Handling: - If framework cannot be determined: STOP and report: "Unable to identify test framework. Please specify test framework or add configuration file."

1. Analyze code to test

- Locate component: - Use codebase_search to find component: "Where is {component} defined or implemented?" - Use grep to search for component name, class name, or function name - If component not found: STOP and report: "Component {component} not found in codebase." - Read component implementation: - Use read_file to read the component file - Understand the structure: functions, methods, classes, exports - Identify inputs and outputs: - Parse function signatures, method parameters - Identify return types and side effects - Note type annotations or JSDoc comments if available - Understand expected behavior: - Analyze implementation logic - Identify business rules and constraints - Review comments or documentation if present - Review existing tests (if any): - Use glob_file_search to find existing test files for the component - Read existing tests to understand coverage and patterns - Note what's already tested vs. what's missing - Identify dependencies to mock: - List external dependencies (imported modules, libraries) - Identify side effects (API calls, database access, file system operations) - Determine what needs to be mocked for isolated unit testing

1. Identify test cases

- Primary Path: Generate from Spec scenarios (if spec was found in Step 1): - For each Gherkin scenario: - Create test case from scenario name - Map Given steps → Arrange (test setup/preconditions) - Map When step → Act (execute component) - Map Then steps → Assert (verify outcomes) - Identify what needs to be mocked based on Given/When steps - Example mapping: - Scenario: "User login with valid credentials" - Given: "User has account with email user@example.com" → Arrange: Mock user database - When: "User submits login form" → Act: Call login(email, password) - Then: "User is authenticated" → Assert: expect(result.authenticated).toBe(true) - Fallback Path: Generate from code analysis (if no spec scenarios): - Happy path (expected inputs): - Normal operation with valid inputs - Typical usage scenarios - Expected return values or behaviors - Edge cases (boundary conditions): - Boundary values (empty strings, zero, max values) - Special values (null, undefined, empty arrays/objects) - Boundary conditions for loops or iterations - Error cases (invalid inputs): - Invalid parameter types - Invalid values (out of range, malformed) - Missing required parameters - Error handling and exception cases - State transitions: - For components with state, test state changes - Initial state, intermediate states, final states - Special values: - Null, undefined, empty collections - Zero, negative numbers (if applicable) - Very large values (if applicable)

1. Generate test code

- Use project's test framework: - Follow framework-specific syntax (describe/it for Jest, test() for Vitest, def test_ for pytest, [Fact] or [Theory] for xUnit, [Test] for NUnit) - Use appropriate assertion library (expect, assert, Assert.Equal, etc.) - Follow naming conventions: - Match existing test file naming pattern - Use descriptive test names that explain what is being tested - Follow existing naming style (camelCase, snake_case, etc.) - Structure tests based on source: - If generated from Spec scenarios (Step 1): - Create describe block: "Spec Contract Validation - {Feature Name}" - For each scenario: - Test name: Use scenario name/description - Add comment linking to spec: // Spec: specs/{FEATURE_DOMAIN}/spec.md - Scenario: {name} - Structure: Arrange (Given) → Act (When) → Assert (Then) - Example: describe('Spec Contract Validation - User Authentication', () => {it('Scenario: User login with valid credentials', () => {// Spec: specs/user-authentication/spec.md - Scenario: User login with valid credentials // Arrange (Given: User has account) const mockUser = {email: 'user@example.com', password: 'hashed'}; mockDb.findUser.mockResolvedValue(mockUser); // Act (When: User submits login) const result = await login('user@example.com', 'password123'); // Assert (Then: User is authenticated) expect(result.authenticated).toBe(true); expect(result.userId).toBe(mockUser.id);});}); - If generated from code analysis (fallback): - Use Arrange-Act-Assert pattern - Group tests by functionality - Use descriptive test names - Include descriptive test names: - Test names should clearly describe what is being tested - Example: "should return user data when valid ID is provided" (not just "test getUser") - Add comments for complex scenarios: - Explain non-obvious setup or assertions - Document why certain mocks or stubs are needed - Adapt mocking strategy for backend vs frontend: - Backend: Mock external services, databases, APIs, file system - Use framework-specific mocking (jest.mock, unittest.mock, Moq, NSubstitute for.NET, etc.) - Mock HTTP clients, database connections, third-party libraries - Frontend: Mock components, hooks/composables, browser APIs, network requests - React: Mock React components using jest.mock or vi.mock, mock custom hooks - Vue 3: Mock Vue components using global.stubs or vi.mock, mock composables with vi.mock, mock child components - Mock browser APIs (fetch, localStorage, window, etc.) using vi.stubGlobal or jest.mock - Use testing library utilities (React Testing Library, Vue Test Utils, etc.)

1. Create test file

- Determine test file location: - Check existing test file organization pattern - Same directory as source (e.g., component.ts → component.test.ts) - Separate test directory (e.g., src/ → tests/ or __tests__/) - Follow project's test file naming convention: - Match existing pattern (e.g., *.test.ts, *_test.py, *.spec.js, *.spec.ts, *Test.cs, *Tests.cs) - Place in correct directory structure - For.NET projects, typically place in separate test project (e.g., Component.Tests/ComponentTests.cs) - For Vue 3 projects with Vitest, commonly uses *.spec.ts or *.spec.js pattern - Match project's test structure: - Follow existing test file template/structure - Use same imports, setup, and teardown patterns - Include same test utilities or helpers if used

1. Run tests

- Execute new tests: - Use run_terminal_cmd to run test command (e.g., npm test, pytest, jest) - Run tests for the specific file or component - Verify all tests pass: - Check test output for pass/fail status - If tests fail: Fix issues and re-run until all pass - Check coverage contribution: - Run coverage command if available (e.g., npm test -- --coverage) - Verify coverage increased for the component - If coverage is low: Add additional test cases for uncovered code paths

1. Review and refine

- Ensure tests are independent: - Tests should not depend on each other - Each test should be able to run in isolation - Clean up any shared state between tests - Verify tests are deterministic: - Tests should produce same results on every run - Remove any randomness or time-dependent behavior - Use fixed values or controlled mocks - Check test clarity and readability: - Test names clearly describe what is being tested - Test code is easy to understand - Comments are added where needed - Add missing test cases if needed: - Review coverage and identify gaps - Add tests for edge cases or error scenarios that were missed - Ensure all public methods/functions are tested

Tools

Codebase Tools

• codebase_search - Find component implementation and related code

- Query: "Where is {component} defined or implemented?" - Query: "How does {component} work?" - Use to locate component and understand usage patterns

• grep - Find specific patterns, functions, or classes

- Pattern: Component names, class names, function names - Use to locate component definitions and usages

• glob_file_search - Find test files and configuration files

- Pattern: **/*.test.*, **/*Test.cs, **/*Tests.cs (test files) - Pattern: **/jest.config.*, **/vitest.config.*, **/pytest.ini, **/*.csproj, **/*.sln (framework configs) - Pattern: **/package.json, **/requirements.txt, **/*.csproj (dependency files) - Use to identify test framework and existing test patterns

Filesystem Tools

• glob_file_search - Find specs, test files, and configuration files

- Pattern: **/specs/*/spec.md (find all specs) - Pattern: **/.test.*, **/*Test.cs, **/*Tests.cs (test files) - Pattern: **/jest.config.*, **/vitest.config.*, **/pytest.ini, **/*.csproj, **/*.sln (framework configs) - Pattern: **/package.json, **/requirements.txt, **/*.csproj (dependency files) - Use to identify specs, test framework, and existing test patterns

• read_file - Read spec Contract scenarios, component code, and existing test files

- Parameters: target_file = path to spec, component, or test file - Use to parse Gherkin scenarios from specs, understand component implementation, and review existing test patterns

• write - Create new test file

- Parameters: file_path = path for new test file, contents = test code - Use to create the test file with generated test code

• list_dir - Explore directory structure

- Parameters: target_directory = directory to explore - Use to understand project structure and locate components

Terminal Tools

• run_terminal_cmd - Execute test commands and check coverage

- Commands: - npm test or npm test -- {test-file} - Run tests (Node.js/Jest/Vitest) - pytest {test-file} - Run tests (Python) - dotnet test or dotnet test --filter {test-file} - Run tests (.NET/xUnit/NUnit/MSTest) - npm test -- --coverage - Run tests with coverage (Node.js) - pytest --cov={module} - Run tests with coverage (Python) - dotnet test --collect:"XPlat Code Coverage" - Run tests with coverage (.NET) - Use to execute tests and verify they pass, check test coverage

Pre-flight Checklist

• Feature domain identified (for spec lookup)
• Spec checked at specs/{FEATURE_DOMAIN}/spec.md
• If spec exists: Contract scenarios parsed
• Component exists in codebase
• Component implementation is readable and understandable
• Codebase structure is understood (project root, source/test directories)
• Test framework identified (Jest, Vitest, pytest, xUnit, NUnit, MSTest, etc.)
• Test file naming convention identified from existing tests
• Test directory structure understood
• Existing test patterns reviewed

Guidance

Role

Act as a QA Engineer or Developer responsible for creating comprehensive unit tests. You are thorough, focused on test quality, and ensure good test coverage while following established patterns and conventions.

Instruction

Execute the create-test workflow to generate unit tests for a specified component. This includes:

1. Checking for Spec and reading Contract scenarios (if spec exists)
2. Detecting the codebase type and test framework
3. Analyzing the component to understand its behavior
4. Identifying comprehensive test cases (from spec scenarios OR from code analysis)
5. Generating well-structured test code following BDD structure (if from spec) or standard patterns
6. Creating the test file and verifying tests pass
7. Ensuring test quality (independence, determinism, clarity)

Context

• Specs may provide Contract scenarios: If spec exists at specs/{FEATURE_DOMAIN}/spec.md, prefer generating tests from Gherkin scenarios
• BDD approach: Gherkin scenarios (Given/When/Then) map to test structure (Arrange/Act/Assert)
• Fallback to code analysis: If no spec exists, use code-driven test generation (current behavior)
• The codebase may use different test frameworks (Jest, Vitest, pytest, unittest, xUnit, NUnit, MSTest, etc.)
• Test files follow project-specific naming conventions and directory structures
• Existing test patterns and utilities should be respected and reused
• Tests should be isolated, fast, and focused on the component being tested
• Backend and frontend code require different mocking strategies
• The goal is to achieve good test coverage while maintaining test quality
• ASDLC patterns: The Spec, Context Gates
• ASDLC pillars: Quality Control (test as output gate), Standardized Parts (Contract scenarios)

Examples

ASDLC: The Spec — Contract scenarios drive tests when present; test output acts as a Context Gates gate.

Example 1: Backend Service Test (Node.js/Jest)

Input: /create-test --type=unit for OAuthService

Component: src/services/auth/OAuthService.ts
Framework: Jest (detected from jest.config.js)
Codebase Type: Backend (Node.js/TypeScript)

Output:
- Test file: src/services/auth/OAuthService.test.ts
- Tests generated:
  - "should return access token when valid credentials provided"
  - "should throw error when invalid credentials provided"
  - "should handle network errors gracefully"
  - "should refresh token when expired"
- Mocks: HTTP client, token storage
- Tests pass, coverage: 85%

Example 2: Frontend Component Test (React/Jest)

Input: /create-test --type=unit for UserProfile

Component: src/components/UserProfile.tsx
Framework: Jest + React Testing Library (detected from package.json)
Codebase Type: Frontend (React/TypeScript)

Output:
- Test file: src/components/UserProfile.test.tsx
- Tests generated:
  - "should render user name and email"
  - "should display loading state when fetching user data"
  - "should handle edit button click"
  - "should display error message when fetch fails"
- Mocks: API calls, user context
- Tests pass, coverage: 80%

Example 2b: Frontend Component Test (Vue 3/Vitest)

Input: /create-test --type=unit for UserProfile

Component: src/components/UserProfile.vue
Framework: Vitest + Vue Test Utils (detected from vitest.config.ts)
Codebase Type: Frontend (Vue 3/TypeScript)

Output:
- Test file: src/components/UserProfile.spec.ts
- Tests generated:
  - "should render user name and email"
  - "should display loading state when fetching user data"
  - "should handle edit button click and emit event"
  - "should display error message when fetch fails"
  - "should update when props change"
- Mocks: API calls, composables, child components
- Uses: mount() from @vue/test-utils, vi.mock for API mocking
- Tests pass, coverage: 82%

Example 2c: BDD Test Generation from Spec (Node.js/Jest)

Input: /create-test --type=unit for OAuthService

Component: src/services/auth/OAuthService.ts
Spec found: specs/user-authentication/spec.md
Framework: Jest (detected from jest.config.js)
Codebase Type: Backend (Node.js/TypeScript)

Spec Contract Scenarios:
- Scenario: User login with valid credentials
- Scenario: Login fails with invalid credentials
- Scenario: Token refresh when expired

Output:
- Test file: src/services/auth/OAuthService.test.ts
- Tests generated FROM SPEC SCENARIOS:
  - "Scenario: User login with valid credentials"
  - "Scenario: Login fails with invalid credentials"
  - "Scenario: Token refresh when expired"
- Test structure:
  - Describe block: "Spec Contract Validation - User Authentication"
  - Each test includes comment: "// Spec: specs/user-authentication/spec.md - Scenario: ..."
  - Arrange (Given) → Act (When) → Assert (Then)
- Mocks: HTTP client, token storage (based on Given steps)
- Tests pass, coverage: 85%

Example 3: Python Function Test (pytest)

Input: /create-test --type=unit for calculate_tax

Component: src/utils/tax.py (function calculate_tax)
Framework: pytest (detected from pytest.ini)
Codebase Type: Backend (Python)

Output:
- Test file: tests/test_tax.py
- Tests generated:
  - "test_calculate_tax_with_valid_amount"
  - "test_calculate_tax_with_zero_amount"
  - "test_calculate_tax_with_negative_amount_raises_error"
  - "test_calculate_tax_with_different_tax_rates"
- Mocks: None (pure function)
- Tests pass, coverage: 100%

Example 4:.NET Service Test (xUnit)

Input: /create-test --type=unit for PaymentService

Component: src/Services/PaymentService.cs
Framework: xUnit (detected from PaymentService.Tests.csproj)
Codebase Type: Backend (.NET/C#)

Output:
- Test file: tests/PaymentService.Tests/PaymentServiceTests.cs
- Tests generated:
  - "ProcessPayment_WithValidCard_ReturnsSuccess"
  - "ProcessPayment_WithInvalidCard_ThrowsPaymentException"
  - "ProcessPayment_WithExpiredCard_ThrowsPaymentException"
  - "ProcessPayment_WithInsufficientFunds_ReturnsDeclined"
- Mocks: IPaymentGateway (using Moq), ILogger
- Tests pass, coverage: 88%

Constraints

Rules (Must Follow):

1. Operational Standards Compliance: This command follows operational standards (documented in AGENTS.md if present, but apply universally):

- File Operations: Follow best practices for file operations and code analysis - Safety Limits: Never commit secrets, API keys, or sensitive data in test files - AGENTS.md Optional: Commands work without AGENTS.md. Standards apply regardless of whether AGENTS.md exists. - See AGENTS.md §3 Operational Boundaries (if present) for detailed standards

1. Prefer Spec Scenarios Over Code Analysis: If spec exists at specs/{FEATURE_DOMAIN}/spec.md, generate tests from Contract scenarios first. Fall back to code analysis only if no spec or no scenarios.
2. BDD Structure for Spec-Driven Tests: When generating from spec scenarios, use BDD structure with describe block "Spec Contract Validation", link to spec in comments, and map Given/When/Then to Arrange/Act/Assert.
3. Component Must Exist: Do not proceed if component cannot be found. STOP and report error.
4. Test Framework Must Be Identifiable: If framework cannot be determined, STOP and ask user.
5. Tests Must Pass: All generated tests must pass. Fix any failures before completing.
6. Follow Existing Patterns: Match existing test file structure, naming, and organization.
7. Test Independence: Each test must be independent and runnable in isolation.
8. Test Determinism: Tests must produce consistent results (no randomness, time-dependent behavior).
9. Comprehensive Coverage: Include all scenarios from spec (if exists), or happy path + edge cases + error cases (if code-driven).
10. Descriptive Test Names: Test names should clearly describe what is being tested. Use scenario names from spec when available.
11. Appropriate Mocking: Mock external dependencies appropriately for backend vs frontend.

Existing Standards (Reference):

• Spec guidance: See specs/README.md for Contract structure and Gherkin scenarios
• Test framework conventions: Follow framework-specific best practices (Jest, Vitest, pytest, xUnit, NUnit, MSTest, etc.)
• Test file organization: Match existing project structure
• Naming conventions: Follow existing test file naming patterns
• Test utilities: Reuse existing test helpers or utilities when available
• ASDLC patterns: BDD (Behavior-Driven Development), Living Specs (tests verify contracts)

Output

1. Test File Created: New test file following project conventions:

- Correct location (matching project structure) - Correct naming (matching naming convention) - Well-structured test code using BDD (if from spec) or AAA pattern (if from code)

1. Tests Generated: Comprehensive test cases covering:

- If from Spec: Test cases for each Gherkin scenario from Contract section - If from code: Happy path scenarios, edge cases, error cases, state transitions - Tests link back to spec (if generated from spec)

1. Tests Pass: All tests execute successfully with no failures
2. Coverage Improved: Test coverage for the component is increased (verify if coverage tools available)

The generated tests should be production-ready, following best practices and project conventions. When spec scenarios are available, tests verify contracts, not just implementation.

Backend vs Frontend Adaptation

Backend Tests

• Mock external services:

- HTTP clients, API calls (use framework mocking like jest.mock, unittest.mock, Moq, NSubstitute for.NET) - Database connections and queries - File system operations - Third-party service integrations

• Test business logic and data transformations:

- Service methods, utility functions - Data validation and processing - Business rule enforcement

• Focus on service methods and API endpoints:

- Test service layer logic - Test API route handlers - Test middleware functions

• Use appropriate backend testing patterns:

- Mock external dependencies - Use test databases or in-memory databases - Test with realistic data structures

Frontend Tests

React Tests

• Mock components, hooks, and browser APIs:

- Mock child components using jest.mock or Vitest vi.mock - Mock custom hooks (React hooks) - Mock browser APIs (fetch, localStorage, window, etc.)

• Test user interactions and state management:

- Component rendering and display - User interactions (clicks, inputs, navigation) - State changes and updates (useState, useEffect, etc.) - Event handling

• Use React Testing Library patterns:

- Focus on testing behavior from user perspective - Use render(), screen.getBy*, fireEvent or userEvent - Avoid testing implementation details

Vue 3 Tests (Vitest + Vue Test Utils)

• Mock components, composables, and browser APIs:

- Mock child components using vi.mock or global.stubs - Mock composables (Vue 3 Composition API) using vi.mock - Mock browser APIs (fetch, localStorage, window, etc.) using vi.stubGlobal

• Test component mounting and rendering:

- Use mount() or shallowMount() from @vue/test-utils - Test component props, emits, slots, and provide/inject - Test Composition API setup (ref, reactive, computed, watch)

• Test user interactions and Vue-specific behavior:

- Component rendering and display - User interactions using wrapper.find() and trigger() - Props updates and reactivity - Event emissions (test emitted events with wrapper.emitted()) - Vue Router navigation (mock router) - Pinia/Vuex state management

• Vue 3 Composition API considerations:

- Test <script setup> components (mount and interact) - Test composables separately if needed - Test reactivity with ref/reactive updates - Mock provide/inject dependencies

• Use appropriate Vue testing patterns:

- Vue Test Utils (mount, shallowMount, Wrapper) - Vitest mocking utilities (vi.mock, vi.fn, vi.spyOn) - Focus on testing component behavior and user interactions - Test props, emits, slots separately from implementation details

Unit Test Checklist

• Feature domain identified (for spec lookup)
• Spec checked at specs/{FEATURE_DOMAIN}/spec.md
• If spec exists: Contract scenarios parsed
• Codebase type detected (backend/frontend)
• Test framework identified
• Component located and analyzed
• Test cases identified:

- From Spec scenarios (if spec exists) - OR From code analysis: Happy path, edge cases, error cases

• Dependencies identified for mocking
• Test file created with correct location and naming
• Tests generated with clear, descriptive names
• Tests follow BDD structure (if from spec) or AAA pattern (if from code)
• If from spec: Tests link to spec in comments
• Tests are independent
• Tests are deterministic
• All tests pass
• Coverage increased (if coverage tools available)