universal-dev-standards 4.1.0 → 4.2.0

package/bundled/locales/zh-TW/integrations/github-copilot/COPILOT-CHAT-REFERENCE.md

@@ -1,8 +1,8 @@
 ---
 source: ../../../../integrations/github-copilot/COPILOT-CHAT-REFERENCE.md
-source_version: 1.0.0
-translation_version: 1.0.0
-last_synced: 2026-01-13
+source_version: 1.1.0
+translation_version: 1.1.0
+last_synced: 2026-01-24
 status: current
 ---
 
@@ -204,6 +204,88 @@ INVEST 檢查清單：
 
 ---
 
+### 9. 重構指導
+
+**Claude Code**：`/refactor`
+
+**Copilot Chat 提示（決策樹）**：
+```
+幫我決定是否應該重構或重寫此程式碼。回答這些問題：
+
+1. 程式碼目前在生產環境運行？[Y/N]
+   → N：考慮重寫（風險較低）
+2. 我理解程式碼做什麼？[Y/N]
+   → N：先寫特徵測試
+3. 測試覆蓋率 > 60%？[Y/N]
+   → N：先加測試
+4. 核心架構可挽救？[Y/N]
+   → N：使用絞殺者無花果模式（漸進替換）
+   → Y：使用漸進式重構
+
+根據我的回答，推薦適當的策略。
+```
+
+**Copilot Chat 提示（戰術重構）**：
+```
+為此程式碼建議戰術重構改進：
+
+**童子軍法則**（小規模、機會性清理）：
+- 重新命名混淆的變數
+- 提取方法
+- 移除死碼
+- 分鐘級，非小時級
+
+**準備性重構**（新增功能前）：
+- 「先讓變更變容易，再做容易的變更」- Kent Beck
+- 重組以適應新需求
+- 與功能工作分開提交
+
+哪種方式適合我的情況？
+```
+
+**Copilot Chat 提示（遺留程式碼安全）**：
+```
+幫我安全地重構此遺留程式碼：
+
+1. **特徵測試**（先做這個）：
+   - 呼叫程式碼，寫預期失敗的斷言
+   - 更新斷言以符合實際行為
+   - 重複直到行為被捕捉
+
+2. **尋找接縫**：
+   - 物件接縫（多型）
+   - 連結接縫（依賴注入）
+
+3. **萌芽/包裝技術**：
+   - 萌芽方法：新邏輯在新的可測試方法中
+   - 包裝方法：在現有程式碼前/後加行為
+
+先幫我寫特徵測試。
+```
+
+**Copilot Chat 提示（戰略重構）**：
+```
+我需要做大規模重構。幫我在以下之間選擇：
+
+1. **絞殺者無花果**：逐步替換遺留系統
+   - 使用時機：替換整個系統，需持續運作
+   - 方法：攔截 → 遷移 → 完成
+
+2. **防腐層 (ACL)**：在新舊之間轉譯
+   - 使用時機：必須與遺留系統共存
+   - 方法：建立 Facade + Adapter + Translator 層
+
+3. **分支抽象**：在主幹上重構共享程式碼
+   - 使用時機：長期重構，無功能分支
+   - 方法：引入抽象 → 加新實作 → 切換 → 移除舊的
+
+哪種模式適合我的情況？我的限制是：[描述情況]
+```
+
+參考：`core/refactoring-standards.md`
+
+---
+
 ## 快速參考卡
 
 | 任務 | Copilot Chat 提示開頭 |
@@ -216,6 +298,9 @@ INVEST 檢查清單：
 | 簽入前 | "在提交前驗證這些品質關卡..." |
 | 發布 | "依照語義化版本準備發布..." |
 | 文件 | "為此撰寫文件..." |
+| 重構 | "幫我決定是否應該重構或重寫..." |
+| 戰術 | "建議戰術重構改進..." |
+| 遺留 | "幫我安全地重構此遺留程式碼..." |
 
 ---
 
@@ -260,6 +345,7 @@ INVEST 檢查清單：
 
 | 版本 | 日期 | 變更 |
 |------|------|------|
+| 1.1.0 | 2026-01-21 | 新增重構指導提示（決策樹、戰術、遺留、戰略） |
 | 1.0.0 | 2026-01-13 | 初始發布 |
 
 ---

package/bundled/locales/zh-TW/integrations/github-copilot/skills-mapping.md

@@ -1,8 +1,8 @@
 ---
 source: ../../../../integrations/github-copilot/skills-mapping.md
-source_version: 1.0.0
-translation_version: 1.0.0
-last_synced: 2026-01-13
+source_version: 1.1.0
+translation_version: 1.1.0
+last_synced: 2026-01-24
 status: current
 ---
 
@@ -36,7 +36,7 @@ Claude Code 提供 18 個 skills 和 16 個斜線命令。GitHub Copilot 不支
 | **git-workflow-guide** | 僅 Chat 提示 | ⚠️ 部分 |
 | **documentation-guide** | 僅 Chat 提示 | ⚠️ 部分 |
 | **methodology-system** | 不可用 | ❌ 無 |
-| **refactoring-assistant** | 僅 Chat 提示 | ⚠️ 部分 |
+| **refactoring-assistant** | COPILOT-CHAT-REFERENCE.md §9 | ✅ 完整 |
 | **error-code-guide** | 僅 Chat 提示 | ⚠️ 部分 |
 | **project-structure-guide** | 僅 Chat 提示 | ⚠️ 部分 |
 | **logging-guide** | 僅 Chat 提示 | ⚠️ 部分 |
@@ -65,6 +65,9 @@ Claude Code 提供 18 個 skills 和 16 個斜線命令。GitHub Copilot 不支
 | `/changelog` | "以 Keep a Changelog 格式產生 CHANGELOG 條目" |
 | `/docs` | "為此函式/模組撰寫文件" |
 | `/spec` | "為此功能建立規格文件" |
+| `/refactor` | "幫我決定是否應該重構或重寫..." |
+| `/refactor tactical` | "建議戰術重構改進..." |
+| `/refactor legacy` | "幫我安全地重構此遺留程式碼..." |
 | `/methodology` | ❌ 不可用 |
 | `/bdd` | "幫我用 Gherkin 格式撰寫 BDD 情境" |
 | `/config` | "為此類型應用程式建議專案結構" |
@@ -183,6 +186,7 @@ gh repo create my-project --template my-org/project-template
 
 | 版本 | 日期 | 變更 |
 |------|------|------|
+| 1.1.0 | 2026-01-21 | 更新 refactoring-assistant 為完整狀態，新增 /refactor 命令對照 |
 | 1.0.0 | 2026-01-13 | 初始發布 |
 
 ---

package/bundled/skills/claude-code/README.md

@@ -81,6 +81,14 @@ These are **triggered by keywords** or specific tasks. Install as Skills:
 
 ## Slash Commands | 斜線命令
 
+> **Claude Code v2.1.3+ Note**: Slash commands and skills are now merged. Both `.claude/commands/review.md` and `.claude/skills/review/SKILL.md` create `/review` and work identically. Your existing `.claude/commands/` files continue to work.
+>
+> **Claude Code v2.1.3+ 注意**：斜線命令與 Skills 已合併。`.claude/commands/review.md` 和 `.claude/skills/review/SKILL.md` 都會創建 `/review`，功能相同。現有的 `.claude/commands/` 檔案繼續運作。
+>
+> **For other AI tools**: OpenCode, Roo Code, Gemini CLI and others still use the traditional commands format.
+>
+> **其他 AI 工具**：OpenCode、Roo Code、Gemini CLI 等仍使用傳統的 commands 格式。
+
 In addition to automatic Skills, this plugin provides **manual slash commands** for explicit actions:
 
 除了自動觸發的 Skills，此插件還提供**手動斜線命令**用於明確的操作：

package/bundled/skills/claude-code/agents/README.md

@@ -0,0 +1,305 @@
+# UDS Agents
+
+> **Language**: English | [繁體中文](../../../locales/zh-TW/skills/claude-code/agents/README.md)
+
+**Version**: 1.1.0
+**Last Updated**: 2026-01-21
+**Status**: Stable
+
+---
+
+## Overview
+
+UDS Agents are specialized AI subagents that can be orchestrated to handle complex development workflows. Unlike Skills (which provide context/knowledge), Agents are autonomous entities that can execute multi-step tasks.
+
+## AGENT.md Format Specification
+
+### Frontmatter Schema
+
+```yaml
+---
+# === REQUIRED FIELDS ===
+name: agent-name              # Unique identifier (kebab-case)
+version: 1.0.0                # Semantic version
+description: |                # Multi-line description for AI matching
+  Brief description of the agent's purpose.
+  Keywords: keyword1, keyword2, keyword3.
+
+# === ROLE CONFIGURATION ===
+role: specialist              # orchestrator | specialist | reviewer
+expertise:                    # Domain expertise areas
+  - system-design
+  - api-design
+  - database-modeling
+
+# === TOOL PERMISSIONS (Claude Code Task tool) ===
+# Specify which tools this agent can use
+allowed-tools:
+  - Read                      # File reading
+  - Glob                      # Pattern matching
+  - Grep                      # Content search
+  - Bash(git:*)               # Git commands only
+  - WebFetch                  # Web fetching
+  - WebSearch                 # Web search
+disallowed-tools:             # Explicitly blocked tools
+  - Write                     # No file writing
+  - Edit                      # No file editing
+
+# === SKILL DEPENDENCIES ===
+# Skills that provide context/knowledge to this agent
+skills:
+  - spec-driven-dev           # Skill name reference
+  - testing-guide
+
+# === MODEL PREFERENCES (Claude Code only) ===
+model: claude-sonnet-4-20250514  # Preferred model
+temperature: 0.3              # Response creativity (0.0-1.0)
+
+# === CONTEXT STRATEGY (RLM-inspired) ===
+# Configuration for handling large codebases and long contexts
+context-strategy:
+  mode: adaptive              # full | chunked | adaptive
+  max-chunk-size: 50000       # Maximum tokens per chunk
+  overlap: 500                # Token overlap between chunks
+  analysis-pattern: hierarchical  # hierarchical | parallel | sequential
+
+# === TRIGGER CONDITIONS ===
+triggers:
+  keywords:                   # Auto-activate on these keywords
+    - architecture
+    - system design
+    - 架構設計
+  commands:                   # Slash commands that invoke this agent
+    - /architect
+---
+```
+
+### Role Types
+
+| Role | Description | Use Case |
+|------|-------------|----------|
+| `orchestrator` | Coordinates multiple agents | Complex workflows, feature development |
+| `specialist` | Deep expertise in specific domain | Architecture, testing, documentation |
+| `reviewer` | Evaluates and provides feedback | Code review, spec review, PR review |
+
+### Tool Permission Patterns
+
+```yaml
+# Full tool access (default if not specified)
+allowed-tools: [*]
+
+# Read-only agent
+allowed-tools: [Read, Glob, Grep]
+disallowed-tools: [Write, Edit, Bash]
+
+# Git-only bash access
+allowed-tools:
+  - Bash(git:*)     # Only git commands
+  - Bash(npm:test)  # Only npm test
+
+# Specific file patterns
+allowed-tools:
+  - Write(*.md)     # Only markdown files
+  - Edit(src/**)    # Only src directory
+```
+
+### Context Strategy Configuration (RLM-inspired)
+
+The `context-strategy` section enables intelligent handling of large codebases and long contexts using RLM (Recursive Language Model) principles.
+
+#### Mode Options
+
+| Mode | Description | Use Case |
+|------|-------------|----------|
+| `full` | Load complete context at once | Small projects, documentation tasks |
+| `chunked` | Divide context into fixed-size chunks | Sequential code review, large file analysis |
+| `adaptive` | Dynamically adjust based on content structure | Complex analysis, architecture exploration |
+
+#### Analysis Patterns
+
+| Pattern | Description | Best For |
+|---------|-------------|----------|
+| `hierarchical` | Analyze high-level structure first, then drill down | Architecture analysis, system design |
+| `parallel` | Process multiple sections simultaneously | Independent module analysis, spec review |
+| `sequential` | Process sections in order, preserving context | Code review, step-by-step analysis |
+
+#### Configuration Examples
+
+```yaml
+# For architecture analysis (need overview first)
+context-strategy:
+  mode: adaptive
+  max-chunk-size: 50000
+  overlap: 500
+  analysis-pattern: hierarchical
+
+# For code review (sequential processing)
+context-strategy:
+  mode: chunked
+  max-chunk-size: 30000
+  overlap: 200
+  analysis-pattern: sequential
+
+# For documentation (need full context)
+context-strategy:
+  mode: full
+  analysis-pattern: hierarchical
+```
+
+### Cross-Tool Execution Modes
+
+| AI Tool | Execution Mode | How It Works |
+|---------|---------------|--------------|
+| Claude Code | `task` | Uses Task tool to spawn independent subagent |
+| OpenCode | `task` | Similar Task tool support |
+| Cursor / Windsurf | `inline` | Injects AGENT.md as context prefix |
+| Copilot / Gemini | `inline` | Converts to prompt snippets |
+
+## Built-in Agents
+
+| Agent | Role | Description |
+|-------|------|-------------|
+| [code-architect](./code-architect.md) | specialist | Software architecture and system design |
+| [test-specialist](./test-specialist.md) | specialist | Testing strategy and test implementation |
+| [reviewer](./reviewer.md) | reviewer | Code review and quality assessment |
+| [doc-writer](./doc-writer.md) | specialist | Documentation and technical writing |
+| [spec-analyst](./spec-analyst.md) | specialist | Specification analysis and requirement extraction |
+
+## Usage
+
+### CLI Installation
+
+```bash
+# List available agents
+uds agent list
+
+# Install specific agent to project
+uds agent install code-architect
+
+# Install all agents
+uds agent install --all
+
+# Install to user directory (global)
+uds agent install code-architect --global
+```
+
+### Direct Invocation in Claude Code
+
+```
+/architect [task description]
+```
+
+Or through natural language triggers:
+
+```
+Please help me design the architecture for a new authentication system.
+```
+
+## Creating Custom Agents
+
+### 1. Create AGENT.md File
+
+```bash
+# In your project
+mkdir -p .claude/agents
+touch .claude/agents/my-agent.md
+```
+
+### 2. Define Frontmatter
+
+```yaml
+---
+name: my-custom-agent
+version: 1.0.0
+description: |
+  Custom agent for specific project needs.
+  Keywords: custom, specific, project.
+
+role: specialist
+expertise: [domain-specific]
+
+allowed-tools: [Read, Glob, Grep, Edit]
+skills: [relevant-skill]
+
+triggers:
+  commands: [/myagent]
+---
+```
+
+### 3. Write Agent Instructions
+
+```markdown
+# My Custom Agent
+
+## Purpose
+
+Describe what this agent does.
+
+## Workflow
+
+1. Step one
+2. Step two
+3. Step three
+
+## Guidelines
+
+- Guideline 1
+- Guideline 2
+```
+
+## Agent vs Skill Comparison
+
+| Aspect | Skill | Agent |
+|--------|-------|-------|
+| **Purpose** | Provide knowledge/context | Execute autonomous tasks |
+| **Execution** | Loaded as context | Spawned as subagent (or inline) |
+| **State** | Stateless | Can maintain task state |
+| **Tool Access** | None (context only) | Configurable permissions |
+| **Triggers** | Manual loading | Keywords, commands, workflows |
+| **Composition** | Referenced by agents | Can use skills as context |
+
+## Integration with Workflows
+
+Agents can be orchestrated through workflow definitions:
+
+```yaml
+# workflows/feature-dev.workflow.yaml
+name: feature-development
+steps:
+  - agent: spec-analyst
+    task: Analyze requirements
+  - agent: code-architect
+    task: Design solution
+  - agent: test-specialist
+    task: Define test strategy
+  - manual: Implementation
+  - agent: reviewer
+    task: Code review
+```
+
+See [workflows/README.md](../workflows/README.md) for workflow documentation.
+
+---
+
+## Related Resources
+
+- [Skills Documentation](../README.md)
+- [Workflows Documentation](../workflows/README.md)
+- [AI Agent Paths Configuration](../../../cli/src/config/ai-agent-paths.js)
+
+---
+
+## Version History
+
+| Version | Date | Changes |
+|---------|------|---------|
+| 1.1.0 | 2026-01-21 | Added RLM-inspired context-strategy configuration |
+| 1.0.0 | 2026-01-20 | Initial release |
+
+---
+
+## License
+
+This documentation is released under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/).
+
+**Source**: [universal-dev-standards](https://github.com/AsiaOstrich/universal-dev-standards)