universal-dev-standards 5.1.0-beta.6 → 5.1.0

package/bundled/locales/zh-TW/core/token-budget.md

@@ -0,0 +1,143 @@
+---
+source: ../../../core/token-budget.md
+source_version: 1.0.0
+translation_version: 1.0.0
+last_synced: 2026-04-20
+status: current
+---
+
+# Token 預算區間標準
+
+> **語言**: [English](../../../core/token-budget.md) | 繁體中文
+
+**版本**: 1.0.0
+**最後更新**: 2026-04-15
+**適用範圍**: 所有有 token 預算限制的 Agent 執行環境
+**Scope**: universal
+**來源**: XSPEC-036（claude-code-book Ch.7 four-zone threshold model）
+
+---
+
+## 目的
+
+Token 閾值四區模型：以使用率百分比劃分四個運作區間，漸進觸發不同強度的保護策略。
+
+比「打到上限才停」更優雅，為使用者提供早期預警和自動降級機會。
+
+---
+
+## 核心規範
+
+- 所有有 token 預算限制的執行環境必須使用四區模型監控使用率
+- WARNING 區必須記錄日誌並發出可觀測事件，不得靜默
+- DANGER 區必須觸發輕量保護策略（如截斷工具結果、縮減輸出預算）
+- BLOCKING 區必須拒絕新請求，回傳 `TOKEN_BUDGET_EXCEEDED` 而非讓請求超時崩潰
+- 壓縮操作本身需保留足夠的輸出空間（否則壓縮本身也可能失敗）
+
+---
+
+## 四區模型
+
+| 區間 | 使用率範圍 | 行動 | 日誌等級 |
+|------|----------|------|---------|
+| **SAFE**（安全） | 0% – 84% | 正常執行 | — |
+| **WARNING**（警告） | 85% – 89% | 發出 `TOKEN_BUDGET_WARNING` 事件，通知 Coordinator／使用者 | `info` |
+| **DANGER**（危險） | 90% – 94% | 觸發輕量壓縮策略 | `warn` |
+| **BLOCKING**（阻塞） | 95% – 100% | 拒絕新請求，回傳 `TOKEN_BUDGET_EXCEEDED` | `error` |
+
+### WARNING 區可選行動
+
+- 降低 `model_tier`（capable → standard）
+- 提示使用者考慮分割任務
+
+### DANGER 區必要行動
+
+- 截斷超大工具結果（Tool Result Snip）
+- 縮減後續 Agent 的 `maxToolRounds`（建議降低 20%）
+
+### DANGER 區可選行動
+
+- 將重要輸出持久化到磁碟，上下文只保留摘要
+
+---
+
+## 閾值常數
+
+| 常數 | 值 |
+|------|-----|
+| `WARNING_THRESHOLD` | `0.85` |
+| `DANGER_THRESHOLD` | `0.90` |
+| `BLOCKING_THRESHOLD` | `0.95` |
+
+---
+
+## 類型定義
+
+### TokenBudgetZone
+
+```
+safe | warning | danger | blocking
+```
+
+### TokenBudgetStatus
+
+| 欄位 | 類型 |
+|------|------|
+| `current_tokens` | `number` |
+| `max_tokens` | `number` |
+| `usage_ratio` | `number` |
+| `zone` | `TokenBudgetZone` |
+
+### TokenBudgetExceededError
+
+| 欄位 | 類型 |
+|------|------|
+| `code` | `"TOKEN_BUDGET_EXCEEDED"` |
+| `current_tokens` | `number` |
+| `max_tokens` | `number` |
+| `usage_ratio` | `number` |
+| `zone` | `"blocking"` |
+
+---
+
+## 壓縮後預算常數參考
+
+| 常數 | 值 |
+|------|-----|
+| `MAX_FILES_TO_RESTORE` | `5` |
+| `TOTAL_TOKEN_BUDGET` | `50000` |
+| `MAX_TOKENS_PER_FILE` | `5000` |
+| `MAX_TOKENS_PER_SKILL` | `5000` |
+
+---
+
+## 遙測事件
+
+**`token_budget_zone_change`**
+
+| 欄位 | 類型 |
+|------|------|
+| `from_zone` | `TokenBudgetZone` |
+| `to_zone` | `TokenBudgetZone` |
+| `usage_ratio` | `number` |
+| `agent_name` | `string` |
+| `timestamp` | `string` |
+
+---
+
+## 適用場景
+
+- DevAP 任務執行（Task token 預算監控）
+- VibeOps 9-Agent Pipeline（跨 Agent 累積上下文監控）
+- VibeOps PipelineMemory Snip 觸發條件
+- 任何有 `maxTotalTokens` 限制的 Agent 執行環境
+
+---
+
+## 錯誤碼
+
+| 代碼 | 說明 |
+|------|------|
+| `TB-001` | `TOKEN_BUDGET_EXCEEDED` — 已進入 BLOCKING 區，拒絕新請求 |
+| `TB-002` | `TOKEN_BUDGET_WARNING` — 已進入 WARNING 區，建議採取行動 |
+| `TB-003` | `SNIP_FAILED` — 輕量壓縮失敗，仍在 DANGER 區 |

package/bundled/locales/zh-TW/core/translation-lifecycle-standards.md

@@ -0,0 +1,159 @@
+---
+source: ../../../core/translation-lifecycle-standards.md
+source_version: 1.0.0
+translation_version: 1.0.0
+last_synced: 2026-04-20
+status: current
+---
+
+# 翻譯生命週期標準
+
+> **語言**: [English](../../../core/translation-lifecycle-standards.md) | 繁體中文
+
+**版本**: 1.0.0
+**最後更新**: 2026-04-20
+**狀態**: Trial（到期 2026-10-20）
+**適用範圍**: 所有具備多語言文件的專案
+**來源**: UDS BUG-A06 事後分析（2026-04-20）
+**父標準**: [documentation-lifecycle](documentation-lifecycle.md)
+
+---
+
+## 目的
+
+翻譯生命週期標準：MISSING 與 OUTDATED 的區別、Semver 嚴重度分級，以及自動化整合（pre-commit hook、release gate）。
+
+`documentation-lifecycle` 標準提到翻譯同步是 release gate 的一環，但未定義如何分類或回應不同程度的漂移。本標準填補這個缺口：翻譯檔案不存在與略微過時有本質差異，major 版本落差與 patch 升版也有本質差異。若不作區分，團隊要麼過度阻塞（任何過時都 fail），要麼阻塞不足（忽略所有過時直到成為用戶可見的問題）。
+
+**證據（BUG-A06 事後分析）**：
+1. UDS 在 3 個月內新增 32 個標準時，因缺乏 MISSING gate，翻譯全部缺失，直到 Q2 audit 才發現。
+2. `anti-hallucination.md` zh-CN 停在 1.5.0，來源已升至 1.5.1——新的 Agent 認識論校準框架段落在 zh-CN 版本中完全缺失，使用者看不到。
+
+---
+
+## 核心規範
+
+- `MISSING`（翻譯檔案不存在）永遠是 release blocker — `exit 1`
+- `MAJOR` 版本落差（來源 X > 翻譯 x，X > x）是 release blocker — `exit 1`
+- `MINOR` 版本落差是 advisory — 醒目警告，不阻塞
+- `PATCH` 版本落差是 advisory — 柔和警告，不阻塞
+- 嚴重度由翻譯 frontmatter 的 `source_version` 與目前來源版本的 semver 比較決定
+- 每個翻譯檔案必須有 YAML frontmatter，包含 `source`、`source_version`、`translation_version`、`last_synced`、`status`
+- 來源標準被修改後，翻譯的 `source_version` 立即過時——這種漂移可在 commit 時透過 pre-commit hook 偵測
+
+---
+
+## 嚴重度分級
+
+| 等級 | 條件 | Exit Code | 行動 |
+|------|------|-----------|------|
+| `MISSING` | 翻譯檔案不存在 | 1 | 發布前建立 |
+| `MAJOR` | 來源 MAJOR > 翻譯 MAJOR | 1 | 正式版發布前更新 |
+| `MINOR` | 來源 MINOR > 翻譯 MINOR（同 MAJOR）| 0 | 下次發布前更新（advisory）|
+| `PATCH` | 來源 PATCH > 翻譯 PATCH（同 MAJOR.MINOR）| 0 | 方便時更新（advisory）|
+| `CURRENT` | source_version == 目前來源版本 | 0 | 無需行動 |
+
+### Semver 差異公式
+
+```
+diff_level = compare(
+  strip_prerelease(current_source_version),
+  strip_prerelease(translation.source_version)
+)
+
+其中：major 不同 → MAJOR，minor 不同 → MINOR，其他 → PATCH
+```
+
+---
+
+## 觸發條件
+
+| 事件 | 必要行動 |
+|------|---------|
+| 新增標準到 `core/` | 在所有支援的語言建立翻譯（MISSING check 阻塞發布）|
+| 標準 PATCH 升版 | 方便時更新翻譯的 `source_version` + `last_synced` |
+| 標準 MINOR 升版（含新段落）| 更新翻譯內容 + frontmatter，下次發布前完成 |
+| 標準 MAJOR 升版（大改寫）| 更新翻譯內容 + frontmatter，當前發布前完成（阻塞）|
+| 手動更新翻譯 | 升版 `translation_version` + `last_synced` |
+
+---
+
+## 翻譯 Frontmatter 協議
+
+每個翻譯檔案必須以以下格式開頭：
+
+```yaml
+---
+source: ../../../core/<filename>.md          # 指向來源的相對路徑
+source_version: <X.Y.Z>                      # 最後同步時的來源版本
+translation_version: <X.Y.Z>                 # 翻譯自身的版本
+last_synced: <YYYY-MM-DD>                    # 最後同步日期
+status: current | outdated | draft           # 人類可讀狀態
+---
+```
+
+更新翻譯時：
+1. 翻譯新增或修改的內容
+2. 設定 `source_version` = 新的來源版本
+3. 設定 `translation_version` = 與 `source_version` 相同（或獨立升版）
+4. 設定 `last_synced` = 今天日期
+5. 設定 `status: current`
+
+---
+
+## 自動化整合
+
+### Pre-Commit Hook
+
+當 `core/*.md` 檔案被暫存時，pre-commit hook 執行 `check-translation-sync.sh` 並顯示 OUTDATED 警告。Hook **永不阻塞** commit（在 commit 時阻塞過於擾人）——純提醒用途。
+
+設定方式：`./scripts/install-hooks.sh`（clone 後執行一次）
+
+### Release Gate（`check-translation-sync.sh`）
+
+在 `npm publish` 前或作為 `pre-release-check.sh` 的一部分執行：
+
+```bash
+bash scripts/check-translation-sync.sh
+# MISSING 或 MAJOR 落差 → exit 1
+# 僅 MINOR/PATCH 落差 → exit 0（附 advisory 輸出）
+```
+
+### Version Bump 整合（`bump-version.sh`）
+
+`bump-version.sh` 在更新版本檔案後自動執行 `check-translation-sync.sh`，在升版時即時顯示翻譯健康狀態快照——讓作者立即知道發布前需要更新什麼。
+
+---
+
+## 情境範例
+
+**情境 1 — 標準 patch 升版（1.0.0 → 1.0.1）**
+- 翻譯 `source_version: 1.0.0`，來源現在是 `1.0.1`
+- 嚴重度：`PATCH` — advisory，exit 0
+- 行動：下次方便時更新，不阻塞發布
+
+**情境 2 — 標準 minor 升版含新段落（1.0.0 → 1.1.0）**
+- 翻譯 `source_version: 1.0.0`，來源現在是 `1.1.0`
+- 嚴重度：`MINOR` — advisory，exit 0
+- 行動：下次發布前更新；zh-CN 使用者缺少新內容
+
+**情境 3 — 標準 major 大改寫（1.x.x → 2.0.0）**
+- 翻譯 `source_version: 1.5.0`，來源現在是 `2.0.0`
+- 嚴重度：`MAJOR` — 阻塞，exit 1
+- 行動：正式版發布前必須更新
+
+**情境 4 — 新標準，無翻譯檔案**
+- `locales/zh-TW/core/new-standard.md` 不存在
+- 嚴重度：`MISSING` — 阻塞，exit 1
+- 行動：發布前建立翻譯檔案
+
+---
+
+## 錯誤碼
+
+| 代碼 | 說明 |
+|------|------|
+| `TRANS-001` | `MISSING_TRANSLATION` — 來源標準的翻譯檔案不存在 |
+| `TRANS-002` | `MAJOR_VERSION_GAP` — 翻譯的 source_version 落後目前來源 MAJOR 版本 |
+| `TRANS-003` | `MISSING_FRONTMATTER` — 翻譯檔案缺少必要的 YAML frontmatter |
+| `TRANS-004` | `STALE_SOURCE_REF` — frontmatter 的 `source` 路徑指向不存在的檔案 |

package/bundled/skills/e2e-assistant/SKILL.md

@@ -53,11 +53,25 @@ Scan coverage gaps between BDD features and existing E2E tests.
 
 ## Supported Frameworks | 支援框架
 
-| Framework | Auto-detected | Template |
-|-----------|:------------:|----------|
-| Playwright | ✅ | `@playwright/test` |
-| Cypress | ✅ | `cy.*` commands |
-| Vitest | ✅ | `describe/it` + async |
+| Ecosystem | Framework | Auto-detected | Detection signal |
+|-----------|-----------|:------------:|----------|
+| JavaScript | Playwright | ✅ | `@playwright/test` in package.json |
+| JavaScript | Cypress | ✅ | `cypress` dep + config file |
+| JavaScript | Vitest | ✅ | `vitest` dep + e2e directory |
+| JavaScript | WebdriverIO | ❌ Manual | listed in SUPPORTED_FRAMEWORKS |
+| Python | pytest-playwright | ✅ | `pytest-playwright` in requirements |
+| Python | Selenium + pytest | ✅ | `selenium` in requirements |
+| Python | Robot Framework | ✅ | `robotframework` in requirements |
+| Python | Behave | ✅ | `behave` in requirements |
+| Go | chromedp | ✅ | `chromedp` in go.mod |
+| Go | rod | ✅ | `rod` in go.mod |
+| Java | Selenium WebDriver | ✅ | `selenium-java` in pom.xml / build.gradle |
+| Java | Playwright for Java | ✅ | `playwright` in pom.xml / build.gradle |
+| Java | Gauge | ✅ | `gauge-java` in pom.xml / build.gradle |
+| Ruby | Capybara | ✅ | `capybara` in Gemfile |
+| Ruby | Watir | ✅ | `watir` in Gemfile |
+| C# | Playwright for .NET | ❌ Manual | not yet auto-detected |
+| C# | Selenium WebDriver | ❌ Manual | not yet auto-detected |
 
 ## Usage | 使用方式
 

package/bundled/skills/testing-guide/SKILL.md

@@ -96,6 +96,7 @@ For complete standards, see:
 - [Testing Standards](../../core/testing-standards.md) - Actionable rules
 - [Testing Theory](./testing-theory.md) - Educational knowledge base
 - [Testing Pyramid](./testing-pyramid.md) - Detailed pyramid ratios
+- [Test Skeleton Templates](./test-skeleton-templates.md) - Multi-language skeletons for UT/IT/ST/Perf/Contract
 
 ### AI-Optimized Format (Token-Efficient)
 
@@ -109,6 +110,10 @@ For AI assistants, use the YAML format files for reduced token usage:
   - Integration Testing: `ai/options/testing/integration-testing.ai.yaml`
   - System Testing: `ai/options/testing/system-testing.ai.yaml`
   - E2E Testing: `ai/options/testing/e2e-testing.ai.yaml`
+  - Security Testing: `ai/options/testing/security-testing.ai.yaml`
+  - Performance Testing: `ai/options/testing/performance-testing.ai.yaml`
+  - Contract Testing: `ai/options/testing/contract-testing.ai.yaml`
+- Skeleton templates (all levels, multi-language): [test-skeleton-templates.md](./test-skeleton-templates.md)
 
 ## Naming Conventions
 

package/bundled/skills/testing-guide/test-skeleton-templates.md

@@ -0,0 +1,316 @@
+# Test Skeleton Templates | 測試骨架模板
+
+> **Language**: English | [繁體中文](../../locales/zh-TW/skills/testing-guide/test-skeleton-templates.md)
+
+**Version**: 2.0.0
+**Last Updated**: 2026-04-14
+
+Language-agnostic test skeleton patterns for all testing pyramid levels.
+AI should read `uds.project.yaml` at runtime to determine the target ecosystem,
+then generate code using the appropriate framework for that language.
+
+---
+
+## How to Use This Document | 如何使用本文件
+
+This document contains **patterns only** — pseudocode that applies to any language.
+
+**AI workflow for generating tests:**
+1. Read `uds.project.yaml` to identify `ecosystem` (e.g. `python`, `go`, `java`, `rust`)
+2. Select the test framework standard for that ecosystem (e.g. pytest, testing, JUnit, #[test])
+3. Apply the pseudocode pattern below, translated into the target language
+4. Fill in `[TODO]` markers with real implementation details
+
+> If `uds.project.yaml` is not present, ask the user for their language/framework before generating.
+
+---
+
+## Unit Test (UT) | 單元測試
+
+**Purpose**: Test a single function/method in isolation — no external dependencies.
+**Speed**: < 100ms per test.
+**Isolation**: Complete (use stubs/mocks for all dependencies).
+
+### Pattern: Arrange-Act-Assert
+
+```
+# [TODO: Replace with target language syntax]
+# Arrange
+subject = new TargetClass()
+input   = <test_value>
+
+# Act
+result = subject.method(input)
+
+# Assert
+assert result == <expected_value>
+```
+
+### Pattern: Error / Edge Case
+
+```
+# [TODO: Replace with target language syntax]
+# Assert that invalid input raises / returns error
+assert throws_error( subject.method(invalid_input) )
+```
+
+### Naming convention
+```
+methodName_scenario_expectedResult
+# Examples:
+#   calculateTotal_withEmptyCart_returnsZero
+#   validateEmail_withInvalidFormat_returnsFalse
+```
+
+### AI generation instruction
+> Read `uds.project.yaml → ecosystem`, then:
+> - Wrap the above pseudocode in the ecosystem's test runner structure
+> - Use the ecosystem's assertion library
+> - Use the ecosystem's mock/stub mechanism for dependencies
+> - Apply the ecosystem's file naming convention (e.g. `_test.go`, `*_spec.rb`, `Test*.java`)
+
+---
+
+## Integration Test (IT) | 整合測試
+
+**Purpose**: Test the boundary between two components (DB, HTTP, queue).
+**Speed**: < 1 second per test.
+**Isolation**: Partial — use real implementations when possible, stubs for external services.
+
+### Pattern: Database Integration
+
+```
+# [TODO: Replace with target language syntax]
+# Arrange — start test database (in-memory or container)
+db   = start_test_database()
+repo = new Repository(db)
+
+# Act
+repo.save(entity)
+found = repo.findById(entity.id)
+
+# Assert
+assert found == entity
+
+# Teardown — rollback or truncate
+db.cleanup()
+```
+
+### Pattern: HTTP API Integration
+
+```
+# [TODO: Replace with target language syntax]
+# Act
+response = http_client.post("/api/resource", body = { key: value })
+
+# Assert
+assert response.status == 201
+assert response.body contains expected_fields
+```
+
+### Pattern: Message Queue Integration
+
+```
+# [TODO: Replace with target language syntax]
+# Act
+publisher.send("topic.created", event)
+received = consumer.receive(timeout = 5s)
+
+# Assert
+assert received.payload == event
+```
+
+### AI generation instruction
+> Read `uds.project.yaml → ecosystem`, then select the appropriate:
+> - In-memory or Testcontainers-compatible DB driver for that ecosystem
+> - HTTP test client for that ecosystem
+> - Embedded queue or mock transport for that ecosystem
+> Never hardcode a specific library — choose based on detected ecosystem.
+
+---
+
+## System Test (ST) | 系統測試
+
+**Purpose**: Test a complete business flow within one subsystem, with external services stubbed.
+**Isolation**: Stub all external HTTP/queue calls; use real DB.
+
+### Pattern: Full Flow with Stubbed External Service
+
+```
+# [TODO: Replace with target language syntax]
+# Arrange — stub external dependency
+stub_server = start_http_stub("/external-api/charge",
+    response = { status: "approved", id: "TX-001" })
+
+# Boot subsystem under test
+app = start_system(config = { external_url: stub_server.url })
+
+# Act
+result = app.run_flow(input_data)
+
+# Assert
+assert result.status == "confirmed"
+assert result.id is not null
+
+# Teardown
+stub_server.stop()
+app.stop()
+```
+
+### AI generation instruction
+> Read `uds.project.yaml → ecosystem`, then select the appropriate:
+> - HTTP stub/mock server for that ecosystem (e.g. WireMock, MSW, responses, httpretty, httpmock)
+> - Application test harness / test client for that framework
+> Never hardcode a specific stub library — choose based on detected ecosystem.
+
+---
+
+## Performance Test | 效能測試
+
+**Purpose**: Validate throughput, latency, and resource usage under load.
+**Focus**: SLOs — p95 latency target, requests-per-second target, error rate threshold.
+
+### Pattern: Load Test
+
+```
+# [TODO: Replace with target load testing tool syntax]
+# Configuration
+rate     = <RPS target>          # [TODO] e.g. 50 requests/sec
+duration = <test duration>       # [TODO] e.g. 60 seconds
+target   = "<API_URL>/api/resource"
+
+# Thresholds
+p95_latency_threshold = <ms>     # [TODO] e.g. 500ms
+error_rate_threshold  = <rate>   # [TODO] e.g. 1%
+
+# Execute
+results = run_load_test(rate, duration, target)
+
+# Assert
+assert results.p95_latency < p95_latency_threshold
+assert results.error_rate  < error_rate_threshold
+```
+
+### AI generation instruction
+> Read `uds.project.yaml → ecosystem`, then select the appropriate load testing tool:
+> - Performance tests are **language-independent by default** (k6, Gatling, Locust, vegeta, wrk)
+> - Prefer the tool already present in the project (check for existing config files)
+> - If none present, suggest the most common tool for the detected ecosystem
+> Set `[TODO]` thresholds based on the project's SLO definition if available.
+
+---
+
+## Contract Test | 合約測試
+
+**Purpose**: Verify API contracts between consumer and provider without deploying both.
+**When**: Microservices, shared APIs, CI on every PR.
+
+### Pattern: Consumer Contract (Consumer-Driven)
+
+```
+# [TODO: Replace with target language / Pact version syntax]
+# Define the contract
+pact = new ConsumerPactBuilder(
+    consumer = "[ConsumerName]",   # [TODO]
+    provider = "[ProviderName]"    # [TODO]
+)
+
+pact.given("resource 123 exists")
+    .upon_receiving("a GET request for resource 123")
+    .with_request(method = GET, path = "/resources/123")
+    .will_respond_with(
+        status = 200,
+        body   = { id: like("123"), name: like("Alice") }  # [TODO] add fields
+    )
+
+# Run consumer test against mock provider
+pact.run_test(lambda mock_url:
+    client = new ApiClient(mock_url)
+    resource = client.get_resource("123")
+    assert resource.id == "123"
+)
+```
+
+### Pattern: Provider Verification
+
+```
+# [TODO: Replace with target language / Pact version syntax]
+# Verify all consumer pacts against the real provider
+verifier = new ProviderVerifier(
+    provider_url  = "http://localhost:[PORT]",  # [TODO]
+    pact_broker   = env("PACT_BROKER_URL"),
+    provider_name = "[ProviderName]",           # [TODO]
+    state_handlers = {
+        "resource 123 exists": lambda: seed_database(id = "123")  # [TODO]
+    }
+)
+verifier.verify()
+```
+
+### Pattern: OpenAPI / Schema Contract Test
+
+```
+# [TODO: Replace with target language syntax]
+# Load OpenAPI spec and validate responses
+spec    = load_openapi_spec("./api/openapi.yaml")  # [TODO] spec path
+server  = start_test_server()
+
+for each critical_endpoint in spec.paths:
+    response = http_client.request(critical_endpoint)
+    assert spec.validates(critical_endpoint, response)
+```
+
+### AI generation instruction
+> Read `uds.project.yaml → ecosystem`, then select:
+> - Consumer-Driven: Pact has SDKs for most languages — use the SDK for the detected ecosystem
+> - OpenAPI: Use the OpenAPI validation library available for the detected ecosystem
+> Never hardcode a specific Pact version or library — choose based on detected ecosystem.
+
+---
+
+## Choosing the Right Test Level | 選擇正確的測試層
+
+```
+Use UT when:
+  ✓ Testing a single function or method
+  ✓ All dependencies can be replaced by stubs
+  ✓ Should run in < 100ms
+
+Use IT when:
+  ✓ Testing interaction with a real DB, HTTP endpoint, or message queue
+  ✓ Crossing a component boundary
+  ✓ Needs a real or containerised dependency
+
+Use ST when:
+  ✓ Testing a complete business flow within one subsystem
+  ✓ External services should be stubbed (not mocked)
+  ✓ Validates non-functional requirements (latency, throughput)
+
+Use E2E when:
+  ✓ Testing a full user journey across multiple systems
+  ✓ Uses a production-like environment
+  ✓ Critical paths only (top 5–10 journeys)
+
+Use Performance when:
+  ✓ Validating SLOs before production
+  ✓ Checking throughput or latency under load
+
+Use Contract when:
+  ✓ Two services share an API contract
+  ✓ You want CI to catch consumer/provider drift early
+```
+
+---
+
+## Related Documents | 相關文件
+
+- [Testing Pyramid](./testing-pyramid.md)
+- [E2E Assistant](../e2e-assistant/SKILL.md)
+- [Contract Test Assistant](../contract-test-assistant/SKILL.md)
+- Core: [testing-standards.md](../../core/testing-standards.md)
+- Options:
+  - [unit-testing.ai.yaml](../../.standards/options/testing/unit-testing.ai.yaml)
+  - [integration-testing.ai.yaml](../../.standards/options/testing/integration-testing.ai.yaml)
+  - [system-testing.ai.yaml](../../.standards/options/testing/system-testing.ai.yaml)
+  - [performance-testing.ai.yaml](../../.standards/options/testing/performance-testing.ai.yaml)
+  - [contract-testing.ai.yaml](../../.standards/options/testing/contract-testing.ai.yaml)

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "universal-dev-standards",
-  "version": "5.1.0-beta.6",
+  "version": "5.1.0",
   "description": "CLI tool for adopting Universal Development Standards",
   "keywords": [
     "documentation",
@@ -55,6 +55,7 @@
     "generate:e2e-spec": "node scripts/generate-e2e-spec.mjs",
     "prepare": "node ../scripts/setup-husky.mjs",
     "test:upgrade": "node scripts/test-upgrade-path.mjs",
+    "check:bundle-parity": "node scripts/check-bundle-parity.mjs",
     "prepack": "node scripts/prepack.mjs"
   },
   "dependencies": {