@dtt_siye/atool 1.5.0 → 1.6.1

package/VERSION

@@ -1 +1 @@
-1.5.0
+1.6.1

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@dtt_siye/atool",
-  "version": "1.5.0",
+  "version": "1.6.1",
   "description": "AI Developer Toolkit - 一键配置 AI IDE 的工具集",
   "bin": {
     "atool": "bin/atool.js"

package/skills/clarify-before-build/SKILL.md

@@ -105,3 +105,36 @@ version: 1.0.0
 - **新功能 + 有测试要求** → `/test-driven-development`
 - **重构 + 有回归风险** → `/systematic-debugging`
 - **完成后** → 检查文档是否需要更新（`/doc-coauthoring`）
+
+## Pre-Implementation Documentation Check
+
+在复杂度评估和澄清问题之后，在路由到 brainstorming 或 writing-plans 之前，**检查文档完整性**：
+
+### 检查清单：项目是否有必要的文档？
+
+- [ ] 检查 `docs/300-requirements/` 是否存在并包含相关 PRD
+- [ ] 检查 `docs/400-architecture/` 是否存在并包含相关架构文档
+- [ ] 检查 `docs/500-design/` 是否存在并包含相关 UI 设计文档
+
+### 缺失时的行为：
+
+告知用户："此功能需要 [PRD/架构文档/UI设计文档]，但尚未创建。建议先创建再进入实施。"
+
+主动提供调用相关技能：
+- 缺少 PRD → 建议 `/requirements-writer`（Quick PRD 或 Full PRD）
+- 缺少架构文档 → 建议 `/software-architecture`
+- 缺少 UI 设计 → 建议 `/ui-ux-pro`（包含 HTML 原型）
+
+### 路由增强：
+
+| 场景 | 文档状态 | 建议行为 |
+|------|---------|---------|
+| Tier 2（新功能）+ 缺少 PRD | `docs/300-requirements/` 为空 | 先调用 `/requirements-writer` |
+| Tier 3（架构级）+ 缺少架构文档 | `docs/400-architecture/` 为空 | 先调用 `/software-architecture` |
+| 任何涉及 UI 的功能 + 缺少 UI 设计 | `docs/500-design/` 为空 | 先调用 `/ui-ux-pro`（含 HTML 原型） |
+
+### 注意事项：
+
+- 此检查适用于**新功能和架构级任务**，不适用于 bug 修复、配置调整、单行改动（Tier 0）
+- 如果用户发出退出信号（"直接做"、"stop asking"等），跳过此检查
+- 已有完整文档的项目无需重复创建

package/skills/development-pipeline/SKILL.md

@@ -0,0 +1,168 @@
+---
+name: development-pipeline
+description: 管理从需求到交付的完整开发生命周期。当用户要求开发新功能、新项目时使用此技能编排全套研发流程。
+version: 1.0.0
+---
+
+# Development Pipeline
+
+完整的开发生命周期编排技能。将所有 aTool 研发技能串联为7个阶段，确保从需求发现到交付的每个步骤都被执行。
+
+## 适用场景
+
+- 用户要求"开发一个新功能"、"做一个新项目"
+- 需要从零开始的完整研发流程
+- 多模块、多步骤的复杂开发任务
+
+**不适用于：**
+- Bug 修复（直接用 `/systematic-debugging`）
+- 配置调整、单行改动（直接执行）
+- 已有完整文档和计划的项目（直接用 `/smart-dispatch` 执行计划）
+
+## 7 阶段流程
+
+### Phase 1: 需求发现
+
+**目标：** 评估复杂度，澄清需求，防止幻觉和偏差。
+
+**技能：** `/clarify-before-build`
+
+- 评估任务复杂度（Tier 0-3）
+- Tier 0: 直接执行，退出流程
+- Tier 1: 问1-2个问题后执行
+- Tier 2: 问3-5个结构化问题
+- Tier 3: 进入 Phase 2
+
+**前置文档检查：** 检查 `docs/300-requirements/`、`docs/400-architecture/`、`docs/500-design/` 是否已存在相关文档。
+
+### Phase 2: 设计
+
+**目标：** 通过协作探索产生设计规范，获得用户批准。
+
+**技能：** `/brainstorming`
+
+- 探索项目上下文
+- 提出2-3个方案及权衡
+- 按复杂度分段展示设计，逐段获得批准
+- 输出：设计规范文档（`docs/superpowers/specs/`）
+
+**注意：** brainstorming 的终态会调用 `/writing-plans`。Phase 3 的文档创建任务会在 writing-plans 的 Enhancement 0 中自动注入。
+
+### Phase 3: 文档化（可在计划中并行执行）
+
+**目标：** 在编号文档目录中创建结构化的专业文档。
+
+根据项目范围，创建以下文档（通过 writing-plans Enhancement 0 的 Task 0.x 自动注入）：
+
+| 文档类型 | 目标目录 | 产出技能 | 触发条件 |
+|---------|---------|---------|---------|
+| PRD（产品需求文档） | `docs/300-requirements/` | `/requirements-writer` | 设计规范包含产品功能、用户故事 |
+| 架构设计文档 | `docs/400-architecture/` | `/software-architecture` | 设计规范包含系统架构、微服务 |
+| UI 设计文档 + HTML 原型 | `docs/500-design/` + `prototypes/` | `/ui-ux-pro`（Workflow Mode） | 设计规范包含 UI 组件、交互 |
+
+**每个文档创建后：**
+1. 用 `/doc-standards-enforcer` 验证（L1 格式 + L2 结构）
+2. 用户审批（特别是 UI 原型需要用户确认）
+3. 单独 commit
+
+### Phase 4: 计划
+
+**目标：** 将设计规范分解为可执行的 TDD 任务。
+
+**技能：** `/writing-plans`（aTool 增强版）
+
+- Enhancement 0: 前置文档阶段（Phase 3 的任务自动在此生成）
+- TDD 任务分解（红-绿-重构循环）
+- Enhancement 1: 后置文档同步任务
+- Enhancement 5: Chunk 分组（支持并行执行）
+
+输出：实施计划文档（`docs/superpowers/plans/`）
+
+### Phase 5: 实施
+
+**目标：** 按计划执行 TDD 开发，每个任务经过审查。
+
+**技能选择（根据计划推荐）：**
+
+| 执行方式 | 技能 | 适用场景 |
+|---------|------|---------|
+| 并行执行 | `/smart-dispatch` | Chunk 化的计划，有 subagent 支持的平台 |
+| 顺序审查 | `/subagent-driven-development` | 需要 per-task 审查的计划 |
+| 单 agent | `/executing-plans` | 无 subagent 支持的平台 |
+
+**每个任务遵循：**
+1. 写失败测试
+2. 验证测试失败
+3. 写最小实现代码
+4. 验证测试通过
+5. 规格审查 + 代码质量审查
+6. Commit
+
+### Phase 6: 验证
+
+**目标：** 用新鲜证据确认完成。
+
+**技能：** `/verification-before-completion`
+
+- 运行所有测试，确认通过
+- 检查文档新鲜度（源文件 mtime vs 文档 mtime）
+- 运行代码质量检查
+- 修复所有 Critical 级别问题
+
+### Phase 7: 交付
+
+**目标：** 决定如何集成开发成果。
+
+**技能：** `/finishing-a-development-branch`
+
+- 验证测试通过（必须门禁）
+- 提供选项：merge 到本地 / 创建 PR / 保留分支 / 丢弃
+- 执行用户选择的操作
+- 清理 worktree
+
+## 流程图
+
+```
+用户请求
+    │
+    ▼
+Phase 1: clarify-before-build (Tier 评估)
+    │
+    ├─ Tier 0/1 → 直接执行，退出流程
+    │
+    ▼ Tier 2/3
+Phase 2: brainstorming (设计规范)
+    │
+    ▼
+Phase 4: writing-plans (计划生成)
+    │
+    ├─ Enhancement 0 自动注入 Phase 3 文档任务:
+    │   ├─ Task 0.1: /requirements-writer → docs/300-requirements/
+    │   ├─ Task 0.2: /software-architecture → docs/400-architecture/
+    │   └─ Task 0.3: /ui-ux-pro → docs/500-design/ + prototypes/
+    │
+    └─ 代码实施任务 (TDD)
+        │
+        ▼
+Phase 5: smart-dispatch / subagent-driven-development
+    │
+    ▼
+Phase 6: verification-before-completion
+    │
+    ▼
+Phase 7: finishing-a-development-branch
+```
+
+## 使用方式
+
+直接调用：
+
+```
+/development-pipeline
+```
+
+或由 clarify-before-build 在 Tier 3 评估后自动路由。
+
+## 与其他技能的关系
+
+本技能是**编排器**，不替代任何子技能的具体逻辑。它定义的是技能之间的**调用顺序和依赖关系**，确保完整流程不被跳过。

package/skills/project-analyze/SKILL.md

@@ -378,7 +378,7 @@ jq '.refine_status' .atool-docs/analysis-state.json
 | 问题 | 诊断命令 | 解决方案 |
 |------|---------|---------|
 | 没有检测到模块 | `cat .atool-docs/pre-scan/manifest.json \| jq '.modules'` | 检查源文件格式或 detect-stack 规则 |
-| phase3_graph 失败 | `cat .atool-docs/analysis-state.json \| jq '.phases.phase3_graph'` | 运行 `source lib/knowledge-graph.sh; assemble_enhanced_graph` 检查错误 |
+| phase3_graph 失败 | `cat .atool-docs/analysis-state.json \| jq '.phases.phase3_graph'` | 运行 `source "$HOME/.claude/skills/project-analyze/lib/knowledge-graph.sh"; assemble_enhanced_graph` 检查错误 |
 | Sub-agent 无法读文件 | `cat .atool-docs/modules/{name}/data.json` | 检查 Phase 2 初稿是否生成 |
 | 精炼 Gate 失败 | `cat atool-analysis/modules/{name}/prd.md \| wc -l` | PRD 文件小于 30 行，需要扩展初稿 |
 

package/skills/project-analyze/phases/phase1-setup.md

@@ -19,7 +19,7 @@
 > PROJECT_ROOT="$(pwd)"
 > DOCS_DIR="$PROJECT_ROOT/.atool-docs"
 > mkdir -p "$DOCS_DIR"
-> source lib/detect-stack.sh 2>/dev/null || true
+> source "$HOME/.claude/skills/project-analyze/lib/detect-stack.sh" 2>/dev/null || source lib/detect-stack.sh 2>/dev/null || true
 > STACK=$(detect_stack "$PROJECT_ROOT" 2>/dev/null || echo "unknown")
 > FILE_COUNT=$(find "$PROJECT_ROOT" -type f \
 >   ! -path "*/node_modules/*" ! -path "*/.git/*" \
@@ -50,7 +50,7 @@
 > PROJECT_ROOT="$(pwd)"
 > DOCS_DIR="$PROJECT_ROOT/.atool-docs"
 > mkdir -p "$DOCS_DIR/pre-scan"
-> source lib/pre-scan.sh 2>/dev/null
+> source "$HOME/.claude/skills/project-analyze/lib/pre-scan.sh" 2>/dev/null || source lib/pre-scan.sh 2>/dev/null || true
 > pre_scan_project "$PROJECT_ROOT" "$DOCS_DIR/pre-scan" 2>/dev/null
 > if [ -f "$DOCS_DIR/pre-scan/manifest.json" ]; then
 >   echo "Pre-scan OK: $(jq '.modules | length' "$DOCS_DIR/pre-scan/manifest.json" 2>/dev/null || echo '?') modules"
@@ -154,8 +154,8 @@
 如果 `lib/compute-importance.sh` 可用，运行重要性评分：
 > **执行命令（可选）：**
 > ```bash
-> source lib/compute-importance.sh 2>/dev/null && \
->   compute_importance_batch "$DOCS_DIR/pre-scan/manifest.json" "$PROJECT_ROOT" 2>/dev/null && \
+> source "$HOME/.claude/skills/project-analyze/lib/compute-importance.sh" 2>/dev/null || source lib/compute-importance.sh 2>/dev/null || true
+> compute_importance_batch "$DOCS_DIR/pre-scan/manifest.json" "$PROJECT_ROOT" 2>/dev/null && \
 >   echo "Importance scoring done" || echo "Importance scoring skipped"
 > ```
 

package/skills/project-analyze/phases/phase3-graph.md

@@ -56,7 +56,7 @@
 > PROJECT_ROOT="$(pwd)"
 > DOCS_DIR="$PROJECT_ROOT/.atool-docs"
 > cd "$PROJECT_ROOT"
-> source lib/knowledge-graph.sh 2>/dev/null
+> source "$HOME/.claude/skills/project-analyze/lib/knowledge-graph.sh" 2>/dev/null || source lib/knowledge-graph.sh 2>/dev/null || true
 > assemble_enhanced_graph "$DOCS_DIR" "$PROJECT_ROOT" 2>/dev/null
 > if [ -f "$DOCS_DIR/knowledge-graph.json" ]; then
 >   NODES=$(jq '.nodes | length' "$DOCS_DIR/knowledge-graph.json" 2>/dev/null || echo 0)
@@ -77,7 +77,7 @@
 > PROJECT_ROOT="$(pwd)"
 > DOCS_DIR="$PROJECT_ROOT/.atool-docs"
 > cd "$PROJECT_ROOT"
-> source lib/multi-dimensional-analysis.sh 2>/dev/null
+> source "$HOME/.claude/skills/project-analyze/lib/multi-dimensional-analysis.sh" 2>/dev/null || source lib/multi-dimensional-analysis.sh 2>/dev/null || true
 > run_multi_dimensional_analysis "$DOCS_DIR" 2>/dev/null
 > if [ -f "$DOCS_DIR/multi-dimensional-analysis.json" ]; then
 >   echo "Multi-dimensional analysis complete"

package/skills/project-analyze/phases/phase5-export.md

@@ -92,11 +92,15 @@
 > ```bash
 > PROJECT_ROOT="$(pwd)"
 > cd "$PROJECT_ROOT"
-> if [ -f "lib/export-analysis.sh" ]; then
+> _SKILL_LIB="$HOME/.claude/skills/project-analyze/lib"
+> if [ -f "$_SKILL_LIB/export-analysis.sh" ]; then
+>   source "$_SKILL_LIB/export-analysis.sh"
+>   register_project "$PROJECT_ROOT"
+> elif [ -f "lib/export-analysis.sh" ]; then
 >   source lib/export-analysis.sh
 >   register_project "$PROJECT_ROOT"
 > else
->   echo "WARNING: lib/export-analysis.sh not found — project not registered in ~/.atool/projects.json"
+>   echo "WARNING: export-analysis.sh not found — project not registered in ~/.atool/projects.json"
 >   echo "Manual registration: add entry to ~/.atool/projects.json"
 > fi
 > ```
@@ -113,7 +117,15 @@
 > # 检查 knowledge-graph.json 是否存在
 > if [ -f "$DOCS_DIR/knowledge-graph.json" ]; then
 >   # 调用 generate-visualization.sh
->   if [ -f "$PROJECT_ROOT/lib/generate-visualization.sh" ]; then
+>   _SKILL_LIB="$HOME/.claude/skills/project-analyze/lib"
+>   if [ -f "$_SKILL_LIB/generate-visualization.sh" ]; then
+>     source "$_SKILL_LIB/generate-visualization.sh"
+>     generate_visualization "$DOCS_DIR"
+>     if [ -f "$DOCS_DIR/visualization/index.html" ]; then
+>       echo "✅ Visualization generated: $DOCS_DIR/visualization/index.html"
+>       du -h "$DOCS_DIR/visualization/index.html" | awk '{print "Size: " $1}'
+>     fi
+>   elif [ -f "$PROJECT_ROOT/lib/generate-visualization.sh" ]; then
 >     source "$PROJECT_ROOT/lib/generate-visualization.sh"
 >     generate_visualization "$DOCS_DIR"
 >     if [ -f "$DOCS_DIR/visualization/index.html" ]; then
@@ -121,7 +133,7 @@
 >       du -h "$DOCS_DIR/visualization/index.html" | awk '{print "Size: " $1}'
 >     fi
 >   else
->     echo "⚠️ lib/generate-visualization.sh not found — skipping visualization"
+>     echo "⚠️ generate-visualization.sh not found — skipping visualization"
 >   fi
 > else
 >   echo "⚠️ knowledge-graph.json not found — cannot generate visualization"

package/skills/ui-ux-pro/SKILL.md

@@ -138,3 +138,72 @@ Use this skill when:
 - Overusing animations
 - Cluttered interfaces
 - Hidden important actions
+
+## Workflow Mode (when invoked by other skills)
+
+When ui-ux-pro is invoked by writing-plans, clarify-before-build, or directly for a UI design task, follow this structured workflow to produce deliverables:
+
+### Step 1: Analyze Requirements
+
+Read the spec/design document to understand UI needs:
+- Identify all pages/screens needed
+- Map user flows and navigation patterns
+- List interactive components (forms, modals, buttons, filters)
+- Identify component states: normal, loading, empty, error, disabled
+- Determine responsive breakpoints needed
+
+### Step 2: Create Design Document
+
+Write a structured UI design document to `docs/500-design/YYYY-MM-DD-<feature-name>.md`:
+
+The document should include:
+- **Page inventory**: All pages with layout descriptions
+- **Component specifications**: Each component with all states (normal/loading/empty/error)
+- **Interaction flows**: Step-by-step user interaction sequences
+- **Responsive behavior**: How layouts adapt across breakpoints
+- **Accessibility requirements**: ARIA labels, keyboard navigation, screen reader support
+- **Design tokens**: Colors, typography, spacing values (matching project theme)
+- **Wireframe descriptions**: Text-based layout descriptions for each page
+
+Validate with `/doc-standards-enforcer` (L1 format + L2 structure).
+
+### Step 3: Create HTML Prototype
+
+Generate an interactive HTML prototype as a single self-contained file at `prototypes/<feature-name>.html`:
+
+The prototype must demonstrate:
+- **All key page layouts** with realistic content (not Lorem ipsum)
+- **Interactive elements**: clickable navigation, working modals, functional forms
+- **Design tokens**: Actual colors, fonts, spacing matching the spec
+- **Responsive behavior**: Works on mobile (375px) and desktop (1280px) widths
+- **Component states**: Show at least normal and error states for forms
+- **Self-contained**: All CSS inline or in `<style>`, no external dependencies
+
+The prototype should be presentable to stakeholders for design review.
+
+### Step 4: User Review
+
+Present the HTML prototype to the user:
+- Open the file in a browser for visual inspection
+- Ask: "Does this design match your expectations? Any changes before implementation?"
+- Wait for approval before proceeding
+
+### Step 5: Commit Deliverables
+
+```bash
+git add docs/500-design/ prototypes/
+git commit -m "docs: add UI design and prototype for <feature name>"
+```
+
+### When to Use Workflow Mode
+
+- Invoked by writing-plans as Task 0.3 (pre-implementation documentation phase)
+- Invoked by clarify-before-build when UI docs are missing
+- User explicitly requests UI design or prototype creation
+- Any new project with UI-facing features
+
+### When NOT to Use Workflow Mode
+
+- Quick design questions ("what color should this button be?")
+- Code review of existing UI (use design review questions above)
+- Small UI tweaks to existing components

package/skills/writing-plans/SKILL.md

@@ -10,6 +10,108 @@ This skill extends `superpowers:writing-plans` with **mandatory documentation sy
 
 **Use the superpowers:writing-plans skill as the base**, then apply these additions:
 
+## Enhancement 0: Pre-Implementation Documentation Phase (Mandatory First Check)
+
+**Before creating ANY code implementation tasks, run this check.** This intercepts the brainstorming → writing-plans flow and ensures structured documents are created in numbered directories before code.
+
+### Step 0: Documentation Completeness Assessment
+
+Read the spec/design document. Determine what types of deliverables this project requires:
+
+| Spec Contains | Required Document | Location | Producing Skill |
+|---------------|------------------|----------|-----------------|
+| Product features, user stories, business requirements | PRD | `docs/300-requirements/` | `/requirements-writer` |
+| System architecture, microservices, data flow, deployment | Architecture Doc | `docs/400-architecture/` | `/software-architecture` |
+| UI components, interactions, page layouts | UI Design Doc + HTML Prototype | `docs/500-design/` | `/ui-ux-pro` |
+
+### Check Procedure:
+
+1. Scan the spec for each category above
+2. Check if the corresponding document already exists in the numbered directory
+3. If MISSING → add a Task 0.x document creation task (see templates below) before any code tasks
+4. If EXISTS → skip, proceed to code tasks
+
+### Task 0.x Templates:
+
+````markdown
+### Task 0.1: Create PRD Document (if spec contains product requirements)
+
+**Files:**
+- Create: `docs/300-requirements/YYYY-MM-DD-<feature-name>.md`
+
+- [ ] **Step 1: Invoke /requirements-writer**
+Use the requirements-writer skill to produce a structured PRD from the design spec's requirements sections.
+
+- [ ] **Step 2: Validate document**
+Invoke /doc-standards-enforcer (L1 format + L2 structure) on the created document.
+
+- [ ] **Step 3: Commit**
+```bash
+git add docs/300-requirements/
+git commit -m "docs: add PRD for <feature name>"
+```
+````
+
+````markdown
+### Task 0.2: Create Architecture Document (if spec contains system architecture)
+
+**Files:**
+- Create: `docs/400-architecture/YYYY-MM-DD-<feature-name>.md`
+
+- [ ] **Step 1: Invoke /software-architecture**
+Use the software-architecture skill to produce a structured architecture document from the design spec's architecture sections.
+
+- [ ] **Step 2: Validate document**
+Invoke /doc-standards-enforcer (L1 format + L2 structure) on the created document.
+
+- [ ] **Step 3: Commit**
+```bash
+git add docs/400-architecture/
+git commit -m "docs: add architecture for <feature name>"
+```
+````
+
+````markdown
+### Task 0.3: Create UI Design Document + HTML Prototype (if spec contains UI/UX design)
+
+**Files:**
+- Create: `docs/500-design/YYYY-MM-DD-<feature-name>.md`
+- Create: `prototypes/<feature-name>.html`
+
+- [ ] **Step 1: Invoke /ui-ux-pro in Workflow Mode**
+Use the ui-ux-pro skill to produce a structured UI design document AND an interactive HTML prototype.
+
+- [ ] **Step 2: Validate document**
+Invoke /doc-standards-enforcer (L1 format + L2 structure) on the design document.
+
+- [ ] **Step 3: Present prototype to user**
+Open the HTML prototype in browser. Get user approval on design before proceeding to code.
+
+- [ ] **Step 4: Commit**
+```bash
+git add docs/500-design/ prototypes/
+git commit -m "docs: add UI design and prototype for <feature name>"
+```
+````
+
+**Only after all applicable Task 0.x tasks are complete, proceed to Chunk 1 (code implementation tasks).**
+
+### Chunk Grouping Update:
+
+When using Enhancement 5 chunk grouping, Task 0.x documents form **Chunk 0: Pre-Implementation Documentation**:
+
+```markdown
+## Chunk 0: Pre-Implementation Documentation
+
+### Task 0.1: Create PRD Document
+### Task 0.2: Create Architecture Document
+### Task 0.3: Create UI Design + HTML Prototype
+
+## Chunk 1: Foundation Tasks
+**Depends on:** Chunk 0
+...
+```
+
 ## Enhancement 1: Documentation Update Task (Mandatory Last Task)
 
 **Every plan MUST include a final task for documentation updates.** This is not optional.
@@ -61,6 +163,8 @@ git commit -m "docs: update documentation for [feature name]"
 
 Add this item to the self-review checklist:
 
+**0. Pre-implementation documentation check:** For specs covering product requirements, system architecture, or UI/UX design, does the plan include Task 0.x documentation creation tasks? If the spec covers these topics and no Task 0.x exists, STOP and add them before proceeding.
+
 **4. Documentation coverage:** For each task that creates or modifies source files, does the final documentation sync task list all affected documentation? Are there code changes with no corresponding doc update planned?
 
 ## Enhancement 3: Task-Level Doc Hints