AgentSkillsCN

testing

全面的测试策略，包括TDD、单元、集成、E2E和属性测试。确保代码质量和防止回归。在编写测试、实现功能测试优先或提高覆盖率时使用。

SKILL.md
--- frontmatter
name: testing
description: Comprehensive testing strategies including TDD, unit, integration, E2E, and property-based testing. Ensures code quality and prevents regressions. Use when writing tests, implementing features test-first, or improving coverage.

Testing

Testing strategies for code quality and regression prevention.

When to Use

  • •Implementing features (test-first)
  • •Fixing bugs (reproduce with test)
  • •Improving test coverage
  • •Refactoring safely

MCP Workflow

code
# 1. Find existing test patterns
serena.search_for_pattern("test|spec|describe", relative_path="test/")

# 2. Check test utilities
serena.get_symbols_overview(relative_path="test/support/")

# 3. Find similar tests
serena.find_symbol("*Test", depth=1, include_body=False)

# 4. Framework testing patterns
context7.get-library-docs("<testing-framework>", topic="testing")

Test Pyramid

code
       /\
      /E2E\        Few, slow, high confidence
     /------\
    /Integration\  Some, medium speed
   /--------------\
  /  Unit Tests    \ Many, fast, focused
 /------------------\

TDD Workflow

1. RED (Failing Test)

code
- Write minimal test for expected behavior
- Run test - verify it FAILS correctly
- Commit: "test: add failing test for [feature]"

2. GREEN (Pass It)

code
- Write minimal code to pass
- Run test - verify it PASSES
- Commit: "feat: implement [feature]"

3. REFACTOR

code
- Improve code structure
- Run tests - must still pass
- Commit: "refactor: improve [component]"

Test Structure (AAA)

code
test "should [behavior] when [condition]":
    # Arrange - setup
    input = create_test_input()
    sut = create_system_under_test()

    # Act - execute
    result = sut.execute(input)

    # Assert - verify
    assert result.status == expected

Test Doubles

TypePurposeUse When
StubCanned responsesPredictable input
MockVerify interactionsVerify calls
SpyRecord + real implBoth needed
FakeWorking test implReal is slow

Mock at boundaries: external APIs, databases, file systems. Don't mock: private methods, value objects, simple utilities.

Bug Fix Pattern

code
1. REPRODUCE: Write failing test
2. VERIFY: Confirm it fails correctly
3. FIX: Implement minimal fix
4. CONFIRM: Test passes
5. EXPAND: Add edge case tests

Output Format

markdown
## Testing: [Scope]

| Category | Count | Status |
|----------|-------|--------|
| Unit | N | PASS/FAIL |
| Integration | M | PASS/FAIL |

### New Tests
1. `test_should_create_when_valid`
2. `test_should_reject_duplicate`

### Issues
- [failures or gaps]

Quality Checklist

  • • Tests are independent (no shared state)
  • • Names describe behavior
  • • Each test verifies one concept
  • • Edge cases covered
  • • No flaky tests