unit-tests-skills

内容来源：README.md（说明文档） · 原始地址 · 查看安装指南

原始内容

Unit Test Skills

A collection of AI agent skills for generating high-quality unit tests. These skills encode battle-tested testing principles that work across any programming language, with additional specialized rules for Java.

Installation

Option 1: Using openskills (Recommended)

openskills automatically generates AGENTS.md for maximum AI agent effectiveness.

# Install skills
npx openskills install clear-solutions/unit-tests-skills

# Auto-generate/update AGENTS.md with installed skills
npx openskills sync

Why openskills? According to Vercel's research, skills alone trigger only 53% of the time. With AGENTS.md, success rate jumps to 100%.

Option 2: Using npx skills

npx skills add clear-solutions/unit-tests-skills

Or install specific skills:

npx skills add clear-solutions/unit-tests-skills --skill generate-test-cases
npx skills add clear-solutions/unit-tests-skills --skill generate-tests

For Claude Code specifically:

npx skills add clear-solutions/unit-tests-skills -a claude-code

Important: After installing with npx skills, manually add the snippet from templates/AGENTS-SNIPPET.md to your project's AGENTS.md file.

Why AGENTS.md Matters

Configuration Success Rate
Skills alone 53%
Skills + prompting 79%
AGENTS.md 100%

AGENTS.md provides persistent context to AI agents on every turn, without requiring them to decide to load skills first. See the full article for details.

Available Skills

Skill Command Description
Generate Tests /generate-tests <target> Full workflow: analyzes code, outputs test cases for review, then generates test code. Supports Java (JUnit 5, Mockito, AssertJ).
Generate Test Cases /generate-test-cases <target> Analysis only: outputs a structured list of test cases in Given-When-Then format without generating code.

Usage

Generate Tests (Primary Skill)

/generate-tests src/services/OrderService.java

This single command handles the full workflow:

  1. Analyzes the source code and outputs a structured list of test cases
  2. Asks you to review the test cases before proceeding
  3. Generates the actual test files
  4. Verifies compilation

Analyze Test Coverage Only

If you only want to see what test cases are needed without generating code:

/generate-test-cases src/services/OrderService.java

Testing Principles

These skills enforce proven testing practices:

General Rules (All Languages)

Rule Description
Test Case Strategy Strict INCLUDE/EXCLUDE criteria - test each code branch, not collection sizes
Naming Conventions {method}_{state}_{outcome} format for clarity
Given-When-Then Clear structure with actual/expected prefixes
Keep Tests Focused One scenario per test, single responsibility
Test Behaviors Test what it does, not how it's implemented
No Logic in Tests KISS > DRY - use literal values, avoid calculations
Clean Test Data Use helpers and builders, never rely on defaults
Cause-Effect Clarity Setup belongs in the test, not in distant @BeforeEach
Public APIs First Test through public interfaces, not private methods
Verify Relevant Args Use any() for irrelevant arguments in mocks

Language-Specific Rules

Java

  • JUnit 5 + Mockito + AssertJ
  • @ExtendWith(MockitoExtension.class) for unit tests
  • FORBIDDEN: @SpringBootTest in unit tests
  • Use ArgumentCaptor instead of any() for DTOs
  • Explicit JSON literals (no objectMapper.writeValueAsString())
  • OutputCaptureExtension for log verification

Project Structure

skills/
├── generate-test-cases/
│   ├── SKILL.md
│   └── rules/general/
└── generate-tests/
    ├── SKILL.md
    └── rules/tests/
        ├── general/
        ├── java/unit/
        └── post-generation/

Test Case Generation Strategy

INCLUDE

  • Each distinct code branch and outcome
  • Each unique return value or exception
  • Separate cases for HTTP 400, 401, 403 (never merge)
  • Negative test cases for validation constraints
  • All paths through private methods (via public API)

EXCLUDE

  • Duplicate scenarios with same observable result
  • Collection size variations (1, 2, 3 items) unless code has explicit size logic
  • Speculative cases (exotic Unicode, massive payloads) unless explicitly handled
  • Null arguments unless parameter is @Nullable
  • Multiple tests for same exception type

Example Output

Test Cases

## Test Cases for OrderService.calculateTotal

### 1. calculateTotal_validProducts_returnsSum
- **Given:** List with products priced at 50.0 and 100.0
- **When:** calculateTotal() is called
- **Then:** Returns 150.0
- **Code branch:** Happy path

### 2. calculateTotal_emptyList_throwsIllegalArgumentException
- **Given:** Empty product list
- **When:** calculateTotal() is called
- **Then:** Throws IllegalArgumentException
- **Code branch:** Validation - empty input

Generated Test

@ExtendWith(MockitoExtension.class)
class OrderServiceTest {

    @Mock
    private ProductRepository productRepository;

    @InjectMocks
    private OrderService orderService;

    @Test
    void calculateTotal_validProducts_returnsSum() {
        // Given
        var product1 = new Product("A", 50.0);
        var product2 = new Product("B", 100.0);
        when(productRepository.findAll()).thenReturn(List.of(product1, product2));

        // When
        double actualTotal = orderService.calculateTotal();

        // Then
        double expectedTotal = 150.0;
        assertThat(actualTotal).isEqualTo(expectedTotal);
    }
}

Contributing

Contributions are welcome via Pull Requests. All PRs require review and approval from maintainers before merging.

When adding new rules:

  1. Place general rules in skills/{skill-name}/rules/general/ (or rules/tests/general/ for generate-tests)
  2. Place language-specific rules in skills/generate-tests/rules/tests/{language}/unit/
  3. Update skill files if new rules need explicit reference
  4. Ensure your changes follow the existing format and style

Repository Protection

This repository uses branch protection rules:

  • Direct pushes to main are disabled
  • All changes require a Pull Request
  • PRs require at least one approval from CODEOWNERS
  • Status checks must pass before merging

Google's unit tests best practices

Legend: Implemented = rule file exists in skills | Not implemented = no rule file yet Priority: HIGH / MEDIUM / LOW — based on impact on unit test quality

Guide to Building Skills for Claude

The Complete Guide to Building Skills for Claude

本页内容由上游仓库自动镜像，如有疑问请访问原始仓库。