locales/zh-TW/skills/testing-guide/SKILL.md
[UDS] 測試金字塔與 UT/IT/ST/E2E 測試撰寫標準
npx skillsauth add asiaostrich/universal-dev-standards testingInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
語言: English | 繁體中文
版本: 1.2.0 最後更新: 2026-01-29 適用範圍: Claude Code Skills
本 Skill 提供測試金字塔標準和系統化測試的最佳實踐,同時支援 ISTQB 和業界通行金字塔框架。
UDS 提供 6 個與測試相關的 Skill。請使用以下決策樹找到合適的工具:
你想做什麼? | What do you want to do?
├── 測量程式碼覆蓋率(行/分支/函式) → /coverage
├── 追蹤哪些需求有測試(AC 可追蹤性) → /ac-coverage
├── 以測試驅動開發進行開發(紅-綠-重構) → /tdd
├── 撰寫 BDD 場景(Given-When-Then) → /bdd
├── 與利害關係人定義驗收測試 → /atdd
└── 學習測試標準與最佳實踐 → /testing(本 Skill)
| Skill | 焦點 | Focus |
|-------|------|-------|
| /testing | 測試標準與最佳實踐參考 | Standards and best practices reference |
| /coverage | 程式碼層級覆蓋率分析 | Code-level coverage analysis |
| /ac-coverage | 需求層級 AC 可追蹤性 | Requirement-level AC traceability |
| /tdd | 紅-綠-重構開發循環 | Red-Green-Refactor development cycle |
| /bdd | Given-When-Then 行為場景 | Behavior scenarios with Given-When-Then |
| /atdd | 與利害關係人定義驗收條件 | Acceptance criteria with stakeholders |
| 框架 | 層級 | 適用場景 | |-----------|--------|----------| | ISTQB | UT → IT/SIT → ST → AT/UAT | 企業級、合規性、正式 QA | | 業界通行金字塔 | UT (70%) → IT (20%) → E2E (10%) | 敏捷、DevOps、CI/CD |
整合測試縮寫說明:
┌─────────┐
│ E2E │ ← 10%(較少、較慢)
─┴─────────┴─
┌─────────────┐
│ IT/SIT │ ← 20%(整合測試)
─┴─────────────┴─
┌─────────────────┐
│ UT │ ← 70%(單元測試)
└─────────────────┘
| 層級 | 範圍 | 速度 | 相依性 | |-------|-------|-------|-------------| | UT | 單一函式/類別 | < 100ms | Mock | | IT/SIT | 元件互動 | 1-10秒 | 真實資料庫(容器化) | | ST | 完整系統(ISTQB) | 分鐘級 | 類生產環境 | | E2E | 使用者旅程 | 30秒+ | 所有真實環境 | | AT/UAT | 業務驗證(ISTQB) | 視情況 | 所有真實環境 |
| 指標 | 最低要求 | 建議值 | |--------|---------|-------------| | 行覆蓋率 | 70% | 85% | | 分支覆蓋率 | 60% | 80% | | 函式覆蓋率 | 80% | 90% |
完整標準請參考:
供 AI 助理使用,請採用 YAML 格式檔案以減少 Token 使用量:
ai/standards/testing.ai.yamlai/options/testing/istqb-framework.ai.yamlai/options/testing/industry-pyramid.ai.yamlai/options/testing/unit-testing.ai.yamlai/options/testing/integration-testing.ai.yamlai/options/testing/system-testing.ai.yamlai/options/testing/e2e-testing.ai.yamlai/options/testing/security-testing.ai.yamlai/options/testing/performance-testing.ai.yamlai/options/testing/contract-testing.ai.yaml[ClassName]Tests.cs # C#
[ClassName].test.ts # TypeScript
[class_name]_test.py # Python
[class_name]_test.go # Go
[MethodName]_[Scenario]_[ExpectedResult]()
should_[behavior]_when_[condition]()
test_[method]_[scenario]_[expected]()
| 類型 | 用途 | 使用時機 | |------|---------|-------------| | Stub | 回傳預定義值 | 固定 API 回應 | | Mock | 驗證互動 | 檢查方法是否被呼叫 | | Fake | 簡化實作 | 記憶體資料庫 | | Spy | 記錄呼叫、委派 | 部分 Mock |
test('method_scenario_expected', () => {
// Arrange - 設定測試資料
const input = createTestInput();
const sut = new SystemUnderTest();
// Act - 執行行為
const result = sut.execute(input);
// Assert - 驗證結果
expect(result).toBe(expected);
});
# === ISTQB FUNDAMENTALS ===
terminology:
error: "Human mistake in thinking"
defect: "Bug in code (caused by error)"
failure: "System behaves incorrectly (caused by defect)"
chain: "Error → Defect → Failure"
oracle_problem:
definition: "How do we know the expected result is correct?"
approaches:
- specification_oracle: "Compare against spec"
- reference_oracle: "Compare against reference impl"
- consistency_oracle: "Same input → same output"
- heuristic_oracle: "Reasonable approximation"
# === STATIC vs DYNAMIC ===
static_testing:
definition: "Examine without executing"
techniques: [reviews, walkthroughs, inspections, static_analysis]
finds: "Defects before runtime"
examples: [ESLint, SonarQube, code_review]
dynamic_testing:
definition: "Execute and observe behavior"
techniques: [unit, integration, system, acceptance]
finds: "Failures during execution"
# === TEST DESIGN TECHNIQUES ===
black_box:
equivalence_partitioning:
principle: "Divide inputs into equivalent classes"
example: "Age: [<0 invalid], [0-17 minor], [18-64 adult], [65+ senior]"
boundary_value:
principle: "Test at boundaries of partitions"
example: "Age: test -1, 0, 17, 18, 64, 65"
decision_table:
principle: "Combinations of conditions → actions"
use: "Complex business rules"
state_transition:
principle: "Valid sequences of states"
use: "Workflow, state machines"
white_box:
statement_coverage: "Every statement executed once"
branch_coverage: "Every decision branch taken"
condition_coverage: "Every condition T/F"
path_coverage: "Every possible path (often impractical)"
# === RISK-BASED TESTING ===
risk_assessment:
likelihood: "How likely to fail?"
impact: "How bad if fails?"
priority: "likelihood × impact"
risk_matrix:
high_high: "Test extensively, first priority"
high_low: "Good coverage"
low_high: "Good coverage"
low_low: "Basic coverage"
# === DEFECT MANAGEMENT ===
defect_lifecycle:
states: [new, assigned, in_progress, fixed, verified, closed]
reopen_trigger: "Verification fails"
severity_vs_priority:
severity: "Technical impact (critical/major/minor/trivial)"
priority: "Business urgency (high/medium/low)"
example: "Typo on login page: low severity, high priority (brand)"
# === TEST ENVIRONMENT ===
isolation_levels:
unit: "In-memory, mocked deps"
integration: "Containerized DB (Docker)"
staging: "Production-like, isolated"
production: "Real, feature flags for testing"
test_data_strategies:
fixtures: "Static predefined data"
factories: "Dynamic generation (faker)"
snapshots: "Sanitized production copy"
synthetic: "Algorithm-generated edge cases"
本 Skill 支援專案特定設定。
CONTRIBUTING.md 的「Disabled Skills」(停用 Skills)區段
CONTRIBUTING.md 的「Testing Standards」(測試標準)區段若未找到設定且上下文不清楚時:
CONTRIBUTING.md 中記錄:## Testing Standards
### Coverage Targets
| Metric | Target |
|--------|--------|
| Line | 80% |
| Branch | 70% |
| Function | 85% |
在專案的 CONTRIBUTING.md 中:
## Testing Standards
### Coverage Targets
| Metric | Target |
|--------|--------|
| Line | 80% |
| Branch | 70% |
| Function | 85% |
### Testing Framework
- Unit Tests: Jest
- Integration Tests: Supertest
- E2E Tests: Playwright
當 /testing 完成後,AI 助理應建議:
測試標準與最佳實踐已掌握。建議下一步 / Testing standards and best practices understood. Suggested next steps:
- 執行
/tdd開始測試驅動開發(紅-綠-重構循環) ⭐ Recommended / 推薦 — 將測試知識立即轉化為實踐 / Turn testing knowledge into practice immediately- 執行
/coverage分析目前程式碼覆蓋率 — 找出測試缺口 / Identify testing gaps- 執行
/bdd撰寫行為驅動的 Given-When-Then 場景 — 從使用者角度定義測試 / Define tests from user perspective
| 版本 | 日期 | 變更內容 | |---------|------|---------| | 1.2.0 | 2026-01-29 | 新增連結至新的 testing-theory.md 知識庫 | | 1.1.0 | 2025-12-29 | 新增測試理論精要 YAML 區段 | | 1.0.0 | 2025-12-24 | 初版:標準區段(目的、相關標準、版本歷史、授權) |
本 Skill 以 CC BY 4.0 授權發布。
來源: universal-dev-standards
development
[UDS] 扫描代码库的调试残留与代码质量问题;可自动修正安全模式。 Use when: before committing, during PR review, or periodic codebase cleanup. Keywords: sweep, debug cleanup, console.log, debugger, TODO, ts-any, code quality, 扫描, 清理.
tools
[UDS] 从规格衍生 BDD 场景、TDD 骨架或 ATDD 表格
development
[UDS] 识别重复流程并以正确的开发深度构建 Skill
tools
[UDS] AI 辅助 git push 安全层:质量门禁 + 协作护栏。 Use when: pushing commits, force pushing, pushing to protected branches, pushing feature branches. Keywords: git push, force push, protected branch, quality gate, push receipt, PR automation, 推送, 保护分支, 质量门禁.