chore: bump version to 6.3.38

- Implemented issue discovery prompt to analyze code from various perspectives (bug, UX, test, quality, security, performance, maintainability, best-practices).
- Created structured issue generation prompt from GitHub URLs or text descriptions, including clarity detection and optional clarification questions.
- Introduced CCW Loop-B hybrid orchestrator pattern for iterative development, featuring a coordinator and specialized workers with batch wait support.
- Defined state management, session structure, and output schemas for the CCW Loop-B workflow.
- Added error handling and best practices documentation for the new features.

.claude/CLAUDE.md

@@ -29,13 +29,18 @@ Available CLI endpoints are dynamically defined by the config file:
   ```
   Bash({ command: "ccw cli -p '...' --tool gemini", run_in_background: true })
   ```
-- **After CLI call**: Stop immediately - let CLI execute in background
+- **After CLI call**: Stop output immediately - let CLI execute in background. **DO NOT use TaskOutput polling** - wait for hook callback to receive results
 
 ### CLI Analysis Calls
 - **Wait for results**: MUST wait for CLI analysis to complete before taking any write action. Do NOT proceed with fixes while analysis is running
 - **Value every call**: Each CLI invocation is valuable and costly. NEVER waste analysis results:
   - Aggregate multiple analysis results before proposing solutions
 
+### CLI Auto-Invoke Triggers
+- **Reference**: See `cli-tools-usage.md` → [Auto-Invoke Triggers](#auto-invoke-triggers) for full specification
+- **Key scenarios**: Self-repair fails, ambiguous requirements, architecture decisions, pattern uncertainty, critical code paths
+- **Principles**: Default `--mode analysis`, no confirmation needed, wait for completion, flexible rule selection
+
 ## Code Diagnostics
 
 - **Prefer `mcp__ide__getDiagnostics`** for code error checking over shell-based TypeScript compilation

.claude/TYPESCRIPT_LSP_SETUP.md

@@ -0,0 +1,366 @@
+# Claude Code TypeScript LSP 配置指南
+
+> 更新日期: 2026-01-20
+> 适用版本: Claude Code v2.0.74+
+
+---
+
+## 目录
+
+1. [方式一：插件市场（推荐）](#方式一插件市场推荐)
+2. [方式二：MCP Server (cclsp)](#方式二mcp-server-cclsp)
+3. [方式三：内置LSP工具](#方式三内置lsp工具)
+4. [配置验证](#配置验证)
+5. [故障排查](#故障排查)
+
+---
+
+## 方式一：插件市场（推荐）
+
+### 步骤 1: 添加插件市场
+
+在Claude Code中执行：
+
+```bash
+/plugin marketplace add boostvolt/claude-code-lsps
+```
+
+### 步骤 2: 安装TypeScript LSP插件
+
+```bash
+# TypeScript/JavaScript支持（推荐vtsls）
+/plugin install vtsls@claude-code-lsps
+```
+
+### 步骤 3: 验证安装
+
+```bash
+/plugin list
+```
+
+应该看到：
+```
+✓ vtsls@claude-code-lsps (enabled)
+✓ pyright-lsp@claude-plugins-official (enabled)
+```
+
+### 配置文件自动更新
+
+安装后，`~/.claude/settings.json` 会自动添加：
+
+```json
+{
+  "enabledPlugins": {
+    "pyright-lsp@claude-plugins-official": true,
+    "vtsls@claude-code-lsps": true
+  }
+}
+```
+
+### 支持的操作
+
+- `goToDefinition` - 跳转到定义
+- `findReferences` - 查找引用
+- `hover` - 显示类型信息
+- `documentSymbol` - 文档符号
+- `getDiagnostics` - 诊断信息
+
+---
+
+## 方式二：MCP Server (cclsp)
+
+### 优势
+
+- **位置容错**：自动修正AI生成的不精确行号
+- **更多功能**：支持重命名、完整诊断
+- **灵活配置**：完全自定义LSP服务器
+
+### 安装步骤
+
+#### 1. 安装TypeScript Language Server
+
+```bash
+npm install -g typescript-language-server typescript
+```
+
+验证安装：
+```bash
+typescript-language-server --version
+```
+
+#### 2. 配置cclsp
+
+运行自动配置：
+```bash
+npx cclsp@latest setup --user
+```
+
+或手动创建配置文件：
+
+**文件位置**: `~/.claude/cclsp.json` 或 `~/.config/claude/cclsp.json`
+
+```json
+{
+  "servers": [
+    {
+      "extensions": ["ts", "tsx", "js", "jsx"],
+      "command": ["typescript-language-server", "--stdio"],
+      "rootDir": ".",
+      "restartInterval": 5,
+      "initializationOptions": {
+        "preferences": {
+          "includeInlayParameterNameHints": "all",
+          "includeInlayPropertyDeclarationTypeHints": true,
+          "includeInlayFunctionParameterTypeHints": true,
+          "includeInlayVariableTypeHints": true
+        }
+      }
+    },
+    {
+      "extensions": ["py", "pyi"],
+      "command": ["pylsp"],
+      "rootDir": ".",
+      "restartInterval": 5
+    }
+  ]
+}
+```
+
+#### 3. 在Claude Code中启用MCP Server
+
+添加到Claude Code配置：
+
+```bash
+# 查看当前MCP配置
+cat ~/.claude/.mcp.json
+
+# 如果没有，创建新的
+```
+
+**文件**: `~/.claude/.mcp.json`
+
+```json
+{
+  "mcpServers": {
+    "cclsp": {
+      "command": "npx",
+      "args": ["cclsp@latest"]
+    }
+  }
+}
+```
+
+### cclsp可用的MCP工具
+
+使用时，Claude Code会自动调用这些工具：
+
+- `find_definition` - 按名称查找定义（支持模糊匹配）
+- `find_references` - 查找所有引用
+- `rename_symbol` - 重命名符号（带备份）
+- `get_diagnostics` - 获取诊断信息
+- `restart_server` - 重启LSP服务器
+
+---
+
+## 方式三：内置LSP工具
+
+### 启用方式
+
+设置环境变量：
+
+**Linux/Mac**:
+```bash
+export ENABLE_LSP_TOOL=1
+claude
+```
+
+**Windows (PowerShell)**:
+```powershell
+$env:ENABLE_LSP_TOOL=1
+claude
+```
+
+**永久启用** (添加到shell配置):
+```bash
+# Linux/Mac
+echo 'export ENABLE_LSP_TOOL=1' >> ~/.bashrc
+source ~/.bashrc
+
+# Windows (PowerShell Profile)
+Add-Content $PROFILE '$env:ENABLE_LSP_TOOL=1'
+```
+
+### 限制
+
+- 需要先安装语言服务器插件（见方式一）
+- 不支持重命名等高级操作
+- 无位置容错功能
+
+---
+
+## 配置验证
+
+### 1. 检查LSP服务器是否可用
+
+```bash
+# 检查TypeScript Language Server
+which typescript-language-server  # Linux/Mac
+where typescript-language-server  # Windows
+
+# 测试运行
+typescript-language-server --stdio
+```
+
+### 2. 在Claude Code中测试
+
+打开任意TypeScript文件，让Claude执行：
+
+```typescript
+// 测试LSP功能
+LSP({
+  operation: "hover",
+  filePath: "path/to/your/file.ts",
+  line: 10,
+  character: 5
+})
+```
+
+### 3. 检查插件状态
+
+```bash
+/plugin list
+```
+
+查看启用的插件：
+```bash
+cat ~/.claude/settings.json | grep enabledPlugins
+```
+
+---
+
+## 故障排查
+
+### 问题 1: "No LSP server available"
+
+**原因**：TypeScript LSP插件未安装或未启用
+
+**解决**：
+```bash
+# 重新安装插件
+/plugin install vtsls@claude-code-lsps
+
+# 检查settings.json
+cat ~/.claude/settings.json
+```
+
+### 问题 2: "typescript-language-server: command not found"
+
+**原因**：未安装TypeScript Language Server
+
+**解决**：
+```bash
+npm install -g typescript-language-server typescript
+
+# 验证
+typescript-language-server --version
+```
+
+### 问题 3: LSP响应慢或超时
+
+**原因**：项目太大或配置不当
+
+**解决**：
+```json
+// 在tsconfig.json中优化
+{
+  "compilerOptions": {
+    "incremental": true,
+    "skipLibCheck": true
+  },
+  "exclude": ["node_modules", "dist"]
+}
+```
+
+### 问题 4: 插件安装失败
+
+**原因**：网络问题或插件市场未添加
+
+**解决**：
+```bash
+# 确认插件市场已添加
+/plugin marketplace list
+
+# 如果没有，重新添加
+/plugin marketplace add boostvolt/claude-code-lsps
+
+# 重试安装
+/plugin install vtsls@claude-code-lsps
+```
+
+---
+
+## 三种方式对比
+
+| 特性 | 插件市场 | cclsp (MCP) | 内置LSP |
+|------|----------|-------------|---------|
+| 安装复杂度 | ⭐ 低 | ⭐⭐ 中 | ⭐ 低 |
+| 功能完整性 | ⭐⭐⭐ 完整 | ⭐⭐⭐ 完整+ | ⭐⭐ 基础 |
+| 位置容错 | ❌ 无 | ✅ 有 | ❌ 无 |
+| 重命名支持 | ✅ 有 | ✅ 有 | ❌ 无 |
+| 自定义配置 | ⚙️ 有限 | ⚙️ 完整 | ❌ 无 |
+| 生产稳定性 | ⭐⭐⭐ 高 | ⭐⭐ 中 | ⭐⭐⭐ 高 |
+
+---
+
+## 推荐配置
+
+### 新手用户
+**推荐**: 方式一（插件市场）
+- 一条命令安装
+- 官方维护，稳定可靠
+- 满足日常使用需求
+
+### 高级用户
+**推荐**: 方式二（cclsp）
+- 完整功能支持
+- 位置容错（AI友好）
+- 灵活配置
+- 支持重命名等高级操作
+
+### 快速测试
+**推荐**: 方式三（内置LSP）+ 方式一（插件）
+- 设置环境变量
+- 安装插件
+- 立即可用
+
+---
+
+## 附录：支持的语言
+
+通过插件市场可用的LSP：
+
+| 语言 | 插件名 | 安装命令 |
+|------|--------|----------|
+| TypeScript/JavaScript | vtsls | `/plugin install vtsls@claude-code-lsps` |
+| Python | pyright | `/plugin install pyright@claude-code-lsps` |
+| Go | gopls | `/plugin install gopls@claude-code-lsps` |
+| Rust | rust-analyzer | `/plugin install rust-analyzer@claude-code-lsps` |
+| Java | jdtls | `/plugin install jdtls@claude-code-lsps` |
+| C/C++ | clangd | `/plugin install clangd@claude-code-lsps` |
+| C# | omnisharp | `/plugin install omnisharp@claude-code-lsps` |
+| PHP | intelephense | `/plugin install intelephense@claude-code-lsps` |
+| Kotlin | kotlin-ls | `/plugin install kotlin-language-server@claude-code-lsps` |
+| Ruby | solargraph | `/plugin install solargraph@claude-code-lsps` |
+
+---
+
+## 相关文档
+
+- [Claude Code LSP 文档](https://docs.anthropic.com/claude-code/lsp)
+- [cclsp GitHub](https://github.com/ktnyt/cclsp)
+- [TypeScript Language Server](https://github.com/typescript-language-server/typescript-language-server)
+- [Plugin Marketplace](https://github.com/boostvolt/claude-code-lsps)
+
+---
+
+**配置完成后，重启Claude Code以应用更改**

.claude/agents/cli-discuss-agent.md

@@ -215,7 +215,7 @@ CONTEXT: @**/* | Memory: {ace_context_summary}
 
 EXPECTED: JSON with feasibility_score, findings, implementation_approaches, technical_concerns, code_locations
 
-RULES: $(cat ~/.claude/workflows/cli-templates/protocols/analysis-protocol.md) |
+CONSTRAINTS:
 - Specific file:line references
 - Quantify effort estimates
 - Concrete pros/cons

.claude/agents/cli-execution-agent.md

@@ -61,6 +61,29 @@ Score = 0
 
 **Extract Keywords**: domains (auth, api, database, ui), technologies (react, typescript, node), actions (implement, refactor, test)
 
+**Plan Context Loading** (when executing from plan.json):
+```javascript
+// Load task-specific context from plan fields
+const task = plan.tasks.find(t => t.id === taskId)
+const context = {
+  // Base context
+  scope: task.scope,
+  modification_points: task.modification_points,
+  implementation: task.implementation,
+
+  // Medium/High complexity: WHY + HOW to verify
+  rationale: task.rationale?.chosen_approach,        // Why this approach
+  verification: task.verification?.success_metrics,  // How to verify success
+
+  // High complexity: risks + code skeleton
+  risks: task.risks?.map(r => r.mitigation),         // Risk mitigations to follow
+  code_skeleton: task.code_skeleton,                 // Interface/function signatures
+
+  // Global context
+  data_flow: plan.data_flow?.diagram                 // Data flow overview
+}
+```
+
 ---
 
 ## Phase 2: Context Discovery
@@ -114,9 +137,10 @@ plan → planning/architecture-planning.txt | planning/task-breakdown.txt
 bug-fix → development/bug-diagnosis.txt
 ```
 
-**3. RULES Field**:
-- Use `$(cat ~/.claude/workflows/cli-templates/prompts/{path}.txt)` directly
-- NEVER escape: `\$`, `\"`, `\'` breaks command substitution
+**3. CONSTRAINTS Field**:
+- Use `--rule <template>` option to auto-load protocol + template (appended to prompt)
+- Template names: `category-function` format (e.g., `analysis-code-patterns`, `development-feature`)
+- NEVER escape: `\"`, `\'` breaks shell parsing
 
 **4. Structured Prompt**:
 ```bash
@@ -125,7 +149,31 @@ TASK: {specific_task_with_details}
 MODE: {analysis|write|auto}
 CONTEXT: {structured_file_references}
 EXPECTED: {clear_output_expectations}
-RULES: $(cat {selected_template}) | {constraints}
+CONSTRAINTS: {constraints}
+```
+
+**5. Plan-Aware Prompt Enhancement** (when executing from plan.json):
+```bash
+# Include rationale in PURPOSE (Medium/High)
+PURPOSE: {task.description}
+  Approach: {task.rationale.chosen_approach}
+  Decision factors: {task.rationale.decision_factors.join(', ')}
+
+# Include code skeleton in TASK (High)
+TASK: {task.implementation.join('\n')}
+  Key interfaces: {task.code_skeleton.interfaces.map(i => i.signature)}
+  Key functions: {task.code_skeleton.key_functions.map(f => f.signature)}
+
+# Include verification in EXPECTED
+EXPECTED: {task.acceptance.join(', ')}
+  Success metrics: {task.verification.success_metrics.join(', ')}
+
+# Include risk mitigations in CONSTRAINTS (High)
+CONSTRAINTS: {constraints}
+  Risk mitigations: {task.risks.map(r => r.mitigation).join('; ')}
+
+# Include data flow context (High)
+Memory: Data flow: {plan.data_flow.diagram}
 ```
 
 ---
@@ -156,8 +204,8 @@ TASK: {task}
 MODE: analysis
 CONTEXT: @**/*
 EXPECTED: {output}
-RULES: $(cat ~/.claude/workflows/cli-templates/prompts/analysis/pattern.txt)
-" --tool gemini --mode analysis --cd {dir}
+CONSTRAINTS: {constraints}
+" --tool gemini --mode analysis --rule analysis-code-patterns --cd {dir}
 
 # Qwen fallback: Replace '--tool gemini' with '--tool qwen'
 ```
@@ -204,11 +252,25 @@ find .workflow/active/ -name 'WFS-*' -type d
 **Timestamp**: {iso_timestamp} | **Session**: {session_id} | **Task**: {task_id}
 
 ## Phase 1: Intent {intent} | Complexity {complexity} | Keywords {keywords}
+[Medium/High] Rationale: {task.rationale.chosen_approach}
+[High] Risks: {task.risks.map(r => `${r.description} → ${r.mitigation}`).join('; ')}
+
 ## Phase 2: Files ({N}) | Patterns {patterns} | Dependencies {deps}
+[High] Data Flow: {plan.data_flow.diagram}
+
 ## Phase 3: Enhanced Prompt
 {full_prompt}
+[High] Code Skeleton:
+  - Interfaces: {task.code_skeleton.interfaces.map(i => i.name).join(', ')}
+  - Functions: {task.code_skeleton.key_functions.map(f => f.signature).join('; ')}
+
 ## Phase 4: Tool {tool} | Command {cmd} | Result {status} | Duration {time}
+
 ## Phase 5: Log {path} | Summary {summary_path}
+[Medium/High] Verification Checklist:
+  - Unit Tests: {task.verification.unit_tests.join(', ')}
+  - Success Metrics: {task.verification.success_metrics.join(', ')}
+
 ## Next Steps: {actions}
 ```
 

.claude/agents/cli-lite-planning-agent.md

@@ -77,6 +77,8 @@ Phase 4: planObject Generation
 
 ## CLI Command Template
 
+### Base Template (All Complexity Levels)
+
 ```bash
 ccw cli -p "
 PURPOSE: Generate plan for {task_description}
@@ -84,12 +86,18 @@ TASK:
 • Analyze task/bug description and context
 • Break down into tasks following schema structure
 • Identify dependencies and execution phases
+• Generate complexity-appropriate fields (rationale, verification, risks, code_skeleton, data_flow)
 MODE: analysis
 CONTEXT: @**/* | Memory: {context_summary}
 EXPECTED:
 ## Summary
 [overview]
 
+## Approach
+[high-level strategy]
+
+## Complexity: {Low|Medium|High}
+
 ## Task Breakdown
 ### T1: [Title] (or FIX1 for fix-plan)
 **Scope**: [module/feature path]
@@ -97,17 +105,54 @@ EXPECTED:
 **Description**: [what]
 **Modification Points**: - [file]: [target] - [change]
 **Implementation**: 1. [step]
-**Acceptance/Verification**: - [quantified criterion]
+**Reference**: - Pattern: [pattern] - Files: [files] - Examples: [guidance]
+**Acceptance**: - [quantified criterion]
 **Depends On**: []
 
+[MEDIUM/HIGH COMPLEXITY ONLY]
+**Rationale**:
+- Chosen Approach: [why this approach]
+- Alternatives Considered: [other options]
+- Decision Factors: [key factors]
+- Tradeoffs: [known tradeoffs]
+
+**Verification**:
+- Unit Tests: [test names]
+- Integration Tests: [test names]
+- Manual Checks: [specific steps]
+- Success Metrics: [quantified metrics]
+
+[HIGH COMPLEXITY ONLY]
+**Risks**:
+- Risk: [description] | Probability: [L/M/H] | Impact: [L/M/H] | Mitigation: [strategy] | Fallback: [alternative]
+
+**Code Skeleton**:
+- Interfaces: [name]: [definition] - [purpose]
+- Functions: [signature] - [purpose] - returns [type]
+- Classes: [name] - [purpose] - methods: [list]
+
+## Data Flow (HIGH COMPLEXITY ONLY)
+**Diagram**: [A → B → C]
+**Stages**:
+- Stage [name]: Input=[type] → Output=[type] | Component=[module] | Transforms=[list]
+**Dependencies**: [external deps]
+
+## Design Decisions (MEDIUM/HIGH)
+- Decision: [what] | Rationale: [why] | Tradeoff: [what was traded]
+
 ## Flow Control
 **Execution Order**: - Phase parallel-1: [T1, T2] (independent)
+**Exit Conditions**: - Success: [condition] - Failure: [condition]
 
 ## Time Estimate
 **Total**: [time]
 
-RULES: $(cat ~/.claude/workflows/cli-templates/prompts/planning/02-breakdown-task-steps.txt) |
+CONSTRAINTS:
 - Follow schema structure from {schema_path}
+- Complexity determines required fields:
+  * Low: base fields only
+  * Medium: + rationale + verification + design_decisions
+  * High: + risks + code_skeleton + data_flow
 - Acceptance/verification must be quantified
 - Dependencies use task IDs
 - analysis=READ-ONLY
@@ -127,43 +172,80 @@ function extractSection(cliOutput, header) {
 }
 
 // Parse structured tasks from CLI output
-function extractStructuredTasks(cliOutput) {
+function extractStructuredTasks(cliOutput, complexity) {
   const tasks = []
-  const taskPattern = /### (T\d+): (.+?)\n\*\*File\*\*: (.+?)\n\*\*Action\*\*: (.+?)\n\*\*Description\*\*: (.+?)\n\*\*Modification Points\*\*:\n((?:- .+?\n)*)\*\*Implementation\*\*:\n((?:\d+\. .+?\n)+)\*\*Reference\*\*:\n((?:- .+?\n)+)\*\*Acceptance\*\*:\n((?:- .+?\n)+)\*\*Depends On\*\*: (.+)/g
+  // Split by task headers
+  const taskBlocks = cliOutput.split(/### (T\d+):/).slice(1)
+
+  for (let i = 0; i < taskBlocks.length; i += 2) {
+    const taskId = taskBlocks[i].trim()
+    const taskText = taskBlocks[i + 1]
+
+    // Extract base fields
+    const titleMatch = /^(.+?)(?=\n)/.exec(taskText)
+    const scopeMatch = /\*\*Scope\*\*: (.+?)(?=\n)/.exec(taskText)
+    const actionMatch = /\*\*Action\*\*: (.+?)(?=\n)/.exec(taskText)
+    const descMatch = /\*\*Description\*\*: (.+?)(?=\n)/.exec(taskText)
+    const depsMatch = /\*\*Depends On\*\*: (.+?)(?=\n|$)/.exec(taskText)
 
-  let match
-  while ((match = taskPattern.exec(cliOutput)) !== null) {
     // Parse modification points
-    const modPoints = match[6].trim().split('\n').filter(s => s.startsWith('-')).map(s => {
-      const m = /- \[(.+?)\]: \[(.+?)\] - (.+)/.exec(s)
-      return m ? { file: m[1], target: m[2], change: m[3] } : null
-    }).filter(Boolean)
-
-    // Parse reference
-    const refText = match[8].trim()
-    const reference = {
-      pattern: (/- Pattern: (.+)/m.exec(refText) || [])[1]?.trim() || "No pattern",
-      files: ((/- Files: (.+)/m.exec(refText) || [])[1] || "").split(',').map(f => f.trim()).filter(Boolean),
-      examples: (/- Examples: (.+)/m.exec(refText) || [])[1]?.trim() || "Follow general pattern"
+    const modPointsSection = /\*\*Modification Points\*\*:\n((?:- .+?\n)*)/.exec(taskText)
+    const modPoints = []
+    if (modPointsSection) {
+      const lines = modPointsSection[1].split('\n').filter(s => s.trim().startsWith('-'))
+      lines.forEach(line => {
+        const m = /- \[(.+?)\]: \[(.+?)\] - (.+)/.exec(line)
+        if (m) modPoints.push({ file: m[1].trim(), target: m[2].trim(), change: m[3].trim() })
+      })
     }
 
-    // Parse depends_on
-    const depsText = match[10].trim()
-    const depends_on = depsText === '[]' ? [] : depsText.replace(/[\[\]]/g, '').split(',').map(s => s.trim()).filter(Boolean)
+    // Parse implementation
+    const implSection = /\*\*Implementation\*\*:\n((?:\d+\. .+?\n)+)/.exec(taskText)
+    const implementation = implSection
+      ? implSection[1].split('\n').map(s => s.replace(/^\d+\. /, '').trim()).filter(Boolean)
+      : []
 
-    tasks.push({
-      id: match[1].trim(),
-      title: match[2].trim(),
-      file: match[3].trim(),
-      action: match[4].trim(),
-      description: match[5].trim(),
+    // Parse reference
+    const refSection = /\*\*Reference\*\*:\n((?:- .+?\n)+)/.exec(taskText)
+    const reference = refSection ? {
+      pattern: (/- Pattern: (.+)/m.exec(refSection[1]) || [])[1]?.trim() || "No pattern",
+      files: ((/- Files: (.+)/m.exec(refSection[1]) || [])[1] || "").split(',').map(f => f.trim()).filter(Boolean),
+      examples: (/- Examples: (.+)/m.exec(refSection[1]) || [])[1]?.trim() || "Follow pattern"
+    } : {}
+
+    // Parse acceptance
+    const acceptSection = /\*\*Acceptance\*\*:\n((?:- .+?\n)+)/.exec(taskText)
+    const acceptance = acceptSection
+      ? acceptSection[1].split('\n').map(s => s.replace(/^- /, '').trim()).filter(Boolean)
+      : []
+
+    const task = {
+      id: taskId,
+      title: titleMatch?.[1].trim() || "Untitled",
+      scope: scopeMatch?.[1].trim() || "",
+      action: actionMatch?.[1].trim() || "Implement",
+      description: descMatch?.[1].trim() || "",
       modification_points: modPoints,
-      implementation: match[7].trim().split('\n').map(s => s.replace(/^\d+\. /, '')).filter(Boolean),
+      implementation,
       reference,
-      acceptance: match[9].trim().split('\n').map(s => s.replace(/^- /, '')).filter(Boolean),
-      depends_on
-    })
+      acceptance,
+      depends_on: depsMatch?.[1] === '[]' ? [] : (depsMatch?.[1] || "").replace(/[\[\]]/g, '').split(',').map(s => s.trim()).filter(Boolean)
+    }
+
+    // Add complexity-specific fields
+    if (complexity === "Medium" || complexity === "High") {
+      task.rationale = extractRationale(taskText)
+      task.verification = extractVerification(taskText)
+    }
+
+    if (complexity === "High") {
+      task.risks = extractRisks(taskText)
+      task.code_skeleton = extractCodeSkeleton(taskText)
+    }
+
+    tasks.push(task)
   }
+
   return tasks
 }
 
@@ -186,14 +268,155 @@ function extractFlowControl(cliOutput) {
   }
 }
 
+// Parse rationale section for a task
+function extractRationale(taskText) {
+  const rationaleMatch = /\*\*Rationale\*\*:\n- Chosen Approach: (.+?)\n- Alternatives Considered: (.+?)\n- Decision Factors: (.+?)\n- Tradeoffs: (.+)/s.exec(taskText)
+  if (!rationaleMatch) return null
+
+  return {
+    chosen_approach: rationaleMatch[1].trim(),
+    alternatives_considered: rationaleMatch[2].split(',').map(s => s.trim()).filter(Boolean),
+    decision_factors: rationaleMatch[3].split(',').map(s => s.trim()).filter(Boolean),
+    tradeoffs: rationaleMatch[4].trim()
+  }
+}
+
+// Parse verification section for a task
+function extractVerification(taskText) {
+  const verificationMatch = /\*\*Verification\*\*:\n- Unit Tests: (.+?)\n- Integration Tests: (.+?)\n- Manual Checks: (.+?)\n- Success Metrics: (.+)/s.exec(taskText)
+  if (!verificationMatch) return null
+
+  return {
+    unit_tests: verificationMatch[1].split(',').map(s => s.trim()).filter(Boolean),
+    integration_tests: verificationMatch[2].split(',').map(s => s.trim()).filter(Boolean),
+    manual_checks: verificationMatch[3].split(',').map(s => s.trim()).filter(Boolean),
+    success_metrics: verificationMatch[4].split(',').map(s => s.trim()).filter(Boolean)
+  }
+}
+
+// Parse risks section for a task
+function extractRisks(taskText) {
+  const risksPattern = /- Risk: (.+?) \| Probability: ([LMH]) \| Impact: ([LMH]) \| Mitigation: (.+?)(?: \| Fallback: (.+?))?(?=\n|$)/g
+  const risks = []
+  let match
+
+  while ((match = risksPattern.exec(taskText)) !== null) {
+    risks.push({
+      description: match[1].trim(),
+      probability: match[2] === 'L' ? 'Low' : match[2] === 'M' ? 'Medium' : 'High',
+      impact: match[3] === 'L' ? 'Low' : match[3] === 'M' ? 'Medium' : 'High',
+      mitigation: match[4].trim(),
+      fallback: match[5]?.trim() || undefined
+    })
+  }
+
+  return risks.length > 0 ? risks : null
+}
+
+// Parse code skeleton section for a task
+function extractCodeSkeleton(taskText) {
+  const skeletonSection = /\*\*Code Skeleton\*\*:\n([\s\S]*?)(?=\n\*\*|$)/.exec(taskText)
+  if (!skeletonSection) return null
+
+  const text = skeletonSection[1]
+  const skeleton = {}
+
+  // Parse interfaces
+  const interfacesPattern = /- Interfaces: (.+?): (.+?) - (.+?)(?=\n|$)/g
+  const interfaces = []
+  let match
+  while ((match = interfacesPattern.exec(text)) !== null) {
+    interfaces.push({ name: match[1].trim(), definition: match[2].trim(), purpose: match[3].trim() })
+  }
+  if (interfaces.length > 0) skeleton.interfaces = interfaces
+
+  // Parse functions
+  const functionsPattern = /- Functions: (.+?) - (.+?) - returns (.+?)(?=\n|$)/g
+  const functions = []
+  while ((match = functionsPattern.exec(text)) !== null) {
+    functions.push({ signature: match[1].trim(), purpose: match[2].trim(), returns: match[3].trim() })
+  }
+  if (functions.length > 0) skeleton.key_functions = functions
+
+  // Parse classes
+  const classesPattern = /- Classes: (.+?) - (.+?) - methods: (.+?)(?=\n|$)/g
+  const classes = []
+  while ((match = classesPattern.exec(text)) !== null) {
+    classes.push({
+      name: match[1].trim(),
+      purpose: match[2].trim(),
+      methods: match[3].split(',').map(s => s.trim()).filter(Boolean)
+    })
+  }
+  if (classes.length > 0) skeleton.classes = classes
+
+  return Object.keys(skeleton).length > 0 ? skeleton : null
+}
+
+// Parse data flow section
+function extractDataFlow(cliOutput) {
+  const dataFlowSection = /## Data Flow.*?\n([\s\S]*?)(?=\n## |$)/.exec(cliOutput)
+  if (!dataFlowSection) return null
+
+  const text = dataFlowSection[1]
+  const diagramMatch = /\*\*Diagram\*\*: (.+?)(?=\n|$)/.exec(text)
+  const depsMatch = /\*\*Dependencies\*\*: (.+?)(?=\n|$)/.exec(text)
+
+  // Parse stages
+  const stagesPattern = /- Stage (.+?): Input=(.+?) → Output=(.+?) \| Component=(.+?)(?: \| Transforms=(.+?))?(?=\n|$)/g
+  const stages = []
+  let match
+  while ((match = stagesPattern.exec(text)) !== null) {
+    stages.push({
+      stage: match[1].trim(),
+      input: match[2].trim(),
+      output: match[3].trim(),
+      component: match[4].trim(),
+      transformations: match[5] ? match[5].split(',').map(s => s.trim()).filter(Boolean) : undefined
+    })
+  }
+
+  return {
+    diagram: diagramMatch?.[1].trim() || null,
+    stages: stages.length > 0 ? stages : undefined,
+    dependencies: depsMatch ? depsMatch[1].split(',').map(s => s.trim()).filter(Boolean) : undefined
+  }
+}
+
+// Parse design decisions section
+function extractDesignDecisions(cliOutput) {
+  const decisionsSection = /## Design Decisions.*?\n([\s\S]*?)(?=\n## |$)/.exec(cliOutput)
+  if (!decisionsSection) return null
+
+  const decisionsPattern = /- Decision: (.+?) \| Rationale: (.+?)(?: \| Tradeoff: (.+?))?(?=\n|$)/g
+  const decisions = []
+  let match
+
+  while ((match = decisionsPattern.exec(decisionsSection[1])) !== null) {
+    decisions.push({
+      decision: match[1].trim(),
+      rationale: match[2].trim(),
+      tradeoff: match[3]?.trim() || undefined
+    })
+  }
+
+  return decisions.length > 0 ? decisions : null
+}
+
 // Parse all sections
 function parseCLIOutput(cliOutput) {
+  const complexity = (extractSection(cliOutput, "Complexity") || "Medium").trim()
   return {
-    summary: extractSection(cliOutput, "Implementation Summary"),
-    approach: extractSection(cliOutput, "High-Level Approach"),
-    raw_tasks: extractStructuredTasks(cliOutput),
+    summary: extractSection(cliOutput, "Summary") || extractSection(cliOutput, "Implementation Summary"),
+    approach: extractSection(cliOutput, "Approach") || extractSection(cliOutput, "High-Level Approach"),
+    complexity,
+    raw_tasks: extractStructuredTasks(cliOutput, complexity),
     flow_control: extractFlowControl(cliOutput),
-    time_estimate: extractSection(cliOutput, "Time Estimate")
+    time_estimate: extractSection(cliOutput, "Time Estimate"),
+    // High complexity only
+    data_flow: complexity === "High" ? extractDataFlow(cliOutput) : null,
+    // Medium/High complexity
+    design_decisions: (complexity === "Medium" || complexity === "High") ? extractDesignDecisions(cliOutput) : null
   }
 }
 ```
@@ -326,7 +549,8 @@ function inferFlowControl(tasks) {
 
 ```javascript
 function generatePlanObject(parsed, enrichedContext, input, schemaType) {
-  const tasks = validateAndEnhanceTasks(parsed.raw_tasks, enrichedContext)
+  const complexity = parsed.complexity || input.complexity || "Medium"
+  const tasks = validateAndEnhanceTasks(parsed.raw_tasks, enrichedContext, complexity)
   assignCliExecutionIds(tasks, input.session.id)  // MANDATORY: Assign CLI execution IDs
   const flow_control = parsed.flow_control?.execution_order?.length > 0 ? parsed.flow_control : inferFlowControl(tasks)
   const focus_paths = [...new Set(tasks.flatMap(t => [t.file || t.scope, ...t.modification_points.map(m => m.file)]).filter(Boolean))]
@@ -338,7 +562,7 @@ function generatePlanObject(parsed, enrichedContext, input, schemaType) {
     flow_control,
     focus_paths,
     estimated_time: parsed.time_estimate || `${tasks.length * 30} minutes`,
-    recommended_execution: (input.complexity === "Low" || input.severity === "Low") ? "Agent" : "Codex",
+    recommended_execution: (complexity === "Low" || input.severity === "Low") ? "Agent" : "Codex",
     _metadata: {
       timestamp: new Date().toISOString(),
       source: "cli-lite-planning-agent",
@@ -348,6 +572,15 @@ function generatePlanObject(parsed, enrichedContext, input, schemaType) {
     }
   }
 
+  // Add complexity-specific top-level fields
+  if (complexity === "Medium" || complexity === "High") {
+    base.design_decisions = parsed.design_decisions || []
+  }
+
+  if (complexity === "High") {
+    base.data_flow = parsed.data_flow || null
+  }
+
   // Schema-specific fields
   if (schemaType === 'fix-plan') {
     return {
@@ -361,10 +594,63 @@ function generatePlanObject(parsed, enrichedContext, input, schemaType) {
     return {
       ...base,
       approach: parsed.approach || "Step-by-step implementation",
-      complexity: input.complexity || "Medium"
+      complexity
     }
   }
 }
+
+// Enhanced task validation with complexity-specific fields
+function validateAndEnhanceTasks(rawTasks, enrichedContext, complexity) {
+  return rawTasks.map((task, idx) => {
+    const enhanced = {
+      id: task.id || `T${idx + 1}`,
+      title: task.title || "Unnamed task",
+      scope: task.scope || task.file || inferFile(task, enrichedContext),
+      action: task.action || inferAction(task.title),
+      description: task.description || task.title,
+      modification_points: task.modification_points?.length > 0
+        ? task.modification_points
+        : [{ file: task.scope || task.file, target: "main", change: task.description }],
+      implementation: task.implementation?.length >= 2
+        ? task.implementation
+        : [`Analyze ${task.scope || task.file}`, `Implement ${task.title}`, `Add error handling`],
+      reference: task.reference || { pattern: "existing patterns", files: enrichedContext.relevant_files.slice(0, 2), examples: "Follow existing structure" },
+      acceptance: task.acceptance?.length >= 1
+        ? task.acceptance
+        : [`${task.title} completed`, `Follows conventions`],
+      depends_on: task.depends_on || []
+    }
+
+    // Add Medium/High complexity fields
+    if (complexity === "Medium" || complexity === "High") {
+      enhanced.rationale = task.rationale || {
+        chosen_approach: "Standard implementation approach",
+        alternatives_considered: [],
+        decision_factors: ["Maintainability", "Performance"],
+        tradeoffs: "None significant"
+      }
+      enhanced.verification = task.verification || {
+        unit_tests: [`test_${task.id.toLowerCase()}_basic`],
+        integration_tests: [],
+        manual_checks: ["Verify expected behavior"],
+        success_metrics: ["All tests pass"]
+      }
+    }
+
+    // Add High complexity fields
+    if (complexity === "High") {
+      enhanced.risks = task.risks || [{
+        description: "Implementation complexity",
+        probability: "Low",
+        impact: "Medium",
+        mitigation: "Incremental development with checkpoints"
+      }]
+      enhanced.code_skeleton = task.code_skeleton || null
+    }
+
+    return enhanced
+  })
+}
 ```
 
 ### Error Handling

.claude/agents/cli-planning-agent.md

@@ -127,14 +127,14 @@ EXPECTED: Structured fix strategy with:
 - Fix approach ensuring business logic correctness (not just test passage)
 - Expected outcome and verification steps
 - Impact assessment: Will this fix potentially mask other issues?
-RULES: $(cat ~/.claude/workflows/cli-templates/prompts/{template}) |
+CONSTRAINTS:
 - For {test_type} tests: {layer_specific_guidance}
 - Avoid 'surgical fixes' that mask underlying issues
 - Provide specific line numbers for modifications
 - Consider previous iteration failures
 - Validate fix doesn't introduce new vulnerabilities
 - analysis=READ-ONLY
-" --tool {cli_tool} --mode analysis --cd {project_root} --timeout {timeout_value}
+" --tool {cli_tool} --mode analysis --rule {template} --cd {project_root} --timeout {timeout_value}
 ```
 
 **Layer-Specific Guidance Injection**:

.claude/agents/code-developer.md

@@ -385,8 +385,12 @@ Before completing any task, verify:
 - Make assumptions - verify with existing code
 - Create unnecessary complexity
 
-**Bash Tool**:
-- Use `run_in_background=false` for all Bash/CLI calls to ensure foreground execution
+**Bash Tool (CLI Execution in Agent)**:
+- Use `run_in_background=false` for all Bash/CLI calls - agent cannot receive task hook callbacks
+- Set timeout ≥60 minutes for CLI commands (hooks don't propagate to subagents):
+  ```javascript
+  Bash(command="ccw cli -p '...' --tool codex --mode write", timeout=3600000)  // 60 min
+  ```
 
 **ALWAYS:**
 - **Search Tool Priority**: ACE (`mcp__ace-tool__search_context`) → CCW (`mcp__ccw-tools__smart_search`) / Built-in (`Grep`, `Glob`, `Read`)

.claude/agents/debug-explore-agent.md

@@ -105,7 +105,7 @@ TASK: • Analyze error pattern • Identify potential root causes • Suggest t
 MODE: analysis
 CONTEXT: @{affected_files}
 EXPECTED: Structured hypothesis list with priority ranking
-RULES: $(cat ~/.claude/workflows/cli-templates/prompts/analysis/01-diagnose-bug-root-cause.txt) | Focus on testable conditions
+CONSTRAINTS: Focus on testable conditions
 " --tool gemini --mode analysis --cd {project_root}
 ```
 
@@ -213,7 +213,7 @@ EXPECTED:
 - Evidence summary
 - Root cause identification (if confirmed)
 - Next steps (if inconclusive)
-RULES: $(cat ~/.claude/workflows/cli-templates/prompts/analysis/01-diagnose-bug-root-cause.txt) | Evidence-based reasoning only
+CONSTRAINTS: Evidence-based reasoning only
 " --tool gemini --mode analysis
 ```
 
@@ -271,7 +271,7 @@ TASK:
 MODE: write
 CONTEXT: @{affected_files}
 EXPECTED: Working fix that addresses root cause
-RULES: $(cat ~/.claude/workflows/cli-templates/prompts/development/02-implement-feature.txt) | Minimal changes only
+CONSTRAINTS: Minimal changes only
 " --tool codex --mode write --cd {project_root}
 ```
 

.claude/agents/doc-generator.md

@@ -70,8 +70,8 @@ The agent supports **two execution modes** based on task JSON's `meta.cli_execut
    CONTEXT: @**/* ./src/modules/auth|code|code:5|dirs:2
    ./src/modules/api|code|code:3|dirs:0
    EXPECTED: Documentation files in .workflow/docs/my_project/src/modules/
-   RULES: $(cat ~/.claude/workflows/cli-templates/prompts/documentation/module-documentation.txt) | Mirror source structure
-   " --tool gemini --mode write --cd src/modules
+   CONSTRAINTS: Mirror source structure
+   " --tool gemini --mode write --rule documentation-module --cd src/modules
    ```
 
 4. **CLI Execution** (Gemini CLI):
@@ -216,7 +216,7 @@ Before completion, verify:
 {
   "step": "analyze_module_structure",
   "action": "Deep analysis of module structure and API",
-  "command": "ccw cli -p \"PURPOSE: Document module comprehensively\nTASK: Extract module purpose, architecture, public API, dependencies\nMODE: analysis\nCONTEXT: @**/* System: [system_context]\nEXPECTED: Complete module analysis for documentation\nRULES: $(cat ~/.claude/workflows/cli-templates/prompts/documentation/module-documentation.txt)\" --tool gemini --mode analysis --cd src/auth",
+  "command": "ccw cli -p \"PURPOSE: Document module comprehensively\nTASK: Extract module purpose, architecture, public API, dependencies\nMODE: analysis\nCONTEXT: @**/* System: [system_context]\nEXPECTED: Complete module analysis for documentation\nCONSTRAINTS: Mirror source structure\" --tool gemini --mode analysis --rule documentation-module --cd src/auth",
   "output_to": "module_analysis",
   "on_error": "fail"
 }

.claude/agents/issue-plan-agent.md

@@ -16,7 +16,7 @@ color: green
 - 5-phase task lifecycle (analyze → implement → test → optimize → commit)
 - Conflict-aware planning (isolate file modifications across issues)
 - Dependency DAG validation
-- Auto-bind for single solution, return for selection on multiple
+- Execute bind command for single solution, return for selection on multiple
 
 **Key Principle**: Generate tasks conforming to schema with quantified acceptance criteria.
 
@@ -56,14 +56,61 @@ Phase 4: Validation & Output (15%)
 ccw issue status <issue-id> --json
 ```
 
-**Step 2**: Analyze and classify
+**Step 2**: Analyze failure history (if present)
+```javascript
+function analyzeFailureHistory(issue) {
+  if (!issue.feedback || issue.feedback.length === 0) {
+    return { has_failures: false };
+  }
+
+  // Extract execution failures
+  const failures = issue.feedback.filter(f => f.type === 'failure' && f.stage === 'execute');
+
+  if (failures.length === 0) {
+    return { has_failures: false };
+  }
+
+  // Parse failure details
+  const failureAnalysis = failures.map(f => {
+    const detail = JSON.parse(f.content);
+    return {
+      solution_id: detail.solution_id,
+      task_id: detail.task_id,
+      error_type: detail.error_type,       // test_failure, compilation, timeout, etc.
+      message: detail.message,
+      stack_trace: detail.stack_trace,
+      timestamp: f.created_at
+    };
+  });
+
+  // Identify patterns
+  const errorTypes = failureAnalysis.map(f => f.error_type);
+  const repeatedErrors = errorTypes.filter((e, i, arr) => arr.indexOf(e) !== i);
+
+  return {
+    has_failures: true,
+    failure_count: failures.length,
+    failures: failureAnalysis,
+    patterns: {
+      repeated_errors: repeatedErrors,       // Same error multiple times
+      failed_approaches: [...new Set(failureAnalysis.map(f => f.solution_id))]
+    }
+  };
+}
+```
+
+**Step 3**: Analyze and classify
 ```javascript
 function analyzeIssue(issue) {
+  const failureAnalysis = analyzeFailureHistory(issue);
+
   return {
     issue_id: issue.id,
     requirements: extractRequirements(issue.context),
     scope: inferScope(issue.title, issue.context),
-    complexity: determineComplexity(issue)  // Low | Medium | High
+    complexity: determineComplexity(issue),  // Low | Medium | High
+    failure_analysis: failureAnalysis,       // Failure context for planning
+    is_replan: failureAnalysis.has_failures  // Flag for replanning
   }
 }
 ```
@@ -104,6 +151,41 @@ mcp__ace-tool__search_context({
 
 #### Phase 3: Solution Planning
 
+**Failure-Aware Planning** (when `issue.failure_analysis.has_failures === true`):
+
+```javascript
+function planWithFailureContext(issue, exploration, failureAnalysis) {
+  // Identify what failed before
+  const failedApproaches = failureAnalysis.patterns.failed_approaches;
+  const rootCauses = failureAnalysis.failures.map(f => ({
+    error: f.error_type,
+    message: f.message,
+    task: f.task_id
+  }));
+
+  // Design alternative approach
+  const approach = `
+    **Previous Attempt Analysis**:
+    - Failed approaches: ${failedApproaches.join(', ')}
+    - Root causes: ${rootCauses.map(r => `${r.error} (${r.task}): ${r.message}`).join('; ')}
+
+    **Alternative Strategy**:
+    - [Describe how this solution addresses root causes]
+    - [Explain what's different from failed approaches]
+    - [Prevention steps to catch same errors earlier]
+  `;
+
+  // Add explicit verification tasks
+  const verificationTasks = rootCauses.map(rc => ({
+    verification_type: rc.error,
+    check: `Prevent ${rc.error}: ${rc.message}`,
+    method: `Add unit test / compile check / timeout limit`
+  }));
+
+  return { approach, verificationTasks };
+}
+```
+
 **Multi-Solution Generation**:
 
 Generate multiple candidate solutions when:
@@ -111,30 +193,30 @@ Generate multiple candidate solutions when:
 - Multiple valid implementation approaches exist
 - Trade-offs between approaches (performance vs simplicity, etc.)
 
-| Condition | Solutions |
-|-----------|-----------|
-| Low complexity, single approach | 1 solution, auto-bind |
-| Medium complexity, clear path | 1-2 solutions |
-| High complexity, multiple approaches | 2-3 solutions, user selection |
+| Condition | Solutions | Binding Action |
+|-----------|-----------|----------------|
+| Low complexity, single approach | 1 solution | Execute bind |
+| Medium complexity, clear path | 1-2 solutions | Execute bind if 1, return if 2+ |
+| High complexity, multiple approaches | 2-3 solutions | Return for selection |
+
+**Binding Decision** (based SOLELY on final `solutions.length`):
+```javascript
+// After generating all solutions
+if (solutions.length === 1) {
+  exec(`ccw issue bind ${issueId} ${solutions[0].id}`);  // MUST execute
+} else {
+  return { pending_selection: solutions };  // Return for user choice
+}
+```
 
 **Solution Evaluation** (for each candidate):
 ```javascript
 {
-  analysis: {
-    risk: "low|medium|high",      // Implementation risk
-    impact: "low|medium|high",    // Scope of changes
-    complexity: "low|medium|high" // Technical complexity
-  },
-  score: 0.0-1.0  // Overall quality score (higher = recommended)
+  analysis: { risk: "low|medium|high", impact: "low|medium|high", complexity: "low|medium|high" },
+  score: 0.0-1.0  // Higher = recommended
 }
 ```
 
-**Selection Flow**:
-1. Generate all candidate solutions
-2. Evaluate and score each
-3. Single solution → auto-bind
-4. Multiple solutions → return `pending_selection` for user choice
-
 **Task Decomposition** following schema:
 ```javascript
 function decomposeTasks(issue, exploration) {
@@ -248,8 +330,8 @@ Write({ file_path: filePath, content: newContent })
 ```
 
 **Step 2: Bind decision**
-- **Single solution** → Auto-bind: `ccw issue bind <issue-id> <solution-id>`
-- **Multiple solutions** → Return for user selection (no bind)
+- 1 solution → Execute `ccw issue bind <issue-id> <solution-id>`
+- 2+ solutions → Return `pending_selection` (no bind)
 
 ---
 
@@ -264,14 +346,7 @@ Write({ file_path: filePath, content: newContent })
 
 Each line is a solution JSON containing tasks. Schema: `cat .claude/workflows/cli-templates/schemas/solution-schema.json`
 
-### 2.2 Binding
-
-| Scenario | Action |
-|----------|--------|
-| Single solution | `ccw issue bind <issue-id> <solution-id>` (auto) |
-| Multiple solutions | Register only, return for selection |
-
-### 2.3 Return Summary
+### 2.2 Return Summary
 
 ```json
 {
@@ -310,15 +385,17 @@ Each line is a solution JSON containing tasks. Schema: `cat .claude/workflows/cl
 **ALWAYS**:
 1. **Search Tool Priority**: ACE (`mcp__ace-tool__search_context`) → CCW (`mcp__ccw-tools__smart_search`) / Built-in (`Grep`, `Glob`, `Read`)
 2. Read schema first: `cat .claude/workflows/cli-templates/schemas/solution-schema.json`
-2. Use ACE semantic search as PRIMARY exploration tool
-3. Fetch issue details via `ccw issue status <id> --json`
-4. Quantify acceptance.criteria with testable conditions
-5. Validate DAG before output
-6. Evaluate each solution with `analysis` and `score`
-7. Write solutions to `.workflow/issues/solutions/{issue-id}.jsonl` (append mode)
-8. For HIGH complexity: generate 2-3 candidate solutions
-9. **Solution ID format**: `SOL-{issue-id}-{uid}` where uid is 4 random alphanumeric chars (e.g., `SOL-GH-123-a7x9`)
-10. **GitHub Reply Task**: If issue has `github_url` or `github_number`, add final task to comment on GitHub issue with completion summary
+3. Use ACE semantic search as PRIMARY exploration tool
+4. Fetch issue details via `ccw issue status <id> --json`
+5. **Analyze failure history**: Check `issue.feedback` for type='failure', stage='execute'
+6. **For replanning**: Reference previous failures in `solution.approach`, add prevention steps
+7. Quantify acceptance.criteria with testable conditions
+8. Validate DAG before output
+9. Evaluate each solution with `analysis` and `score`
+10. Write solutions to `.workflow/issues/solutions/{issue-id}.jsonl` (append mode)
+11. For HIGH complexity: generate 2-3 candidate solutions
+12. **Solution ID format**: `SOL-{issue-id}-{uid}` where uid is 4 random alphanumeric chars (e.g., `SOL-GH-123-a7x9`)
+13. **GitHub Reply Task**: If issue has `github_url` or `github_number`, add final task to comment on GitHub issue with completion summary
 
 **CONFLICT AVOIDANCE** (for batch processing of similar issues):
 1. **File isolation**: Each issue's solution should target distinct files when possible
@@ -332,9 +409,9 @@ Each line is a solution JSON containing tasks. Schema: `cat .claude/workflows/cl
 2. Use vague criteria ("works correctly", "good performance")
 3. Create circular dependencies
 4. Generate more than 10 tasks per issue
-5. **Bind when multiple solutions exist** - MUST check `solutions.length === 1` before calling `ccw issue bind`
+5. Skip bind when `solutions.length === 1` (MUST execute bind command)
 
 **OUTPUT**:
-1. Write solutions to `.workflow/issues/solutions/{issue-id}.jsonl` (JSONL format)
-2. Single solution → `ccw issue bind <issue-id> <solution-id>`; Multiple → return only
-3. Return JSON with `bound`, `pending_selection`
+1. Write solutions to `.workflow/issues/solutions/{issue-id}.jsonl`
+2. Execute bind or return `pending_selection` based on solution count
+3. Return JSON: `{ bound: [...], pending_selection: [...] }`

.claude/agents/issue-queue-agent.md

@@ -87,7 +87,7 @@ TASK: • Detect file conflicts (same file modified by multiple solutions)
 MODE: analysis
 CONTEXT: @.workflow/issues/solutions/**/*.jsonl | Solution data: \${SOLUTIONS_JSON}
 EXPECTED: JSON array of conflicts with type, severity, solutions, recommended_order
-RULES: $(cat ~/.claude/workflows/cli-templates/protocols/analysis-protocol.md) | Severity: high (API/data) > medium (file/dependency) > low (architecture)
+CONSTRAINTS: Severity: high (API/data) > medium (file/dependency) > low (architecture)
 " --tool gemini --mode analysis --cd .workflow/issues
 ```
 

.claude/commands/cli/codex-review.md

@@ -0,0 +1,361 @@
+---
+name: codex-review
+description: Interactive code review using Codex CLI via ccw endpoint with configurable review target, model, and custom instructions
+argument-hint: "[--uncommitted|--base <branch>|--commit <sha>] [--model <model>] [--title <title>] [prompt]"
+allowed-tools: Bash(*), AskUserQuestion(*), Read(*)
+---
+
+# Codex Review Command (/cli:codex-review)
+
+## Overview
+Interactive code review command that invokes `codex review` via ccw cli endpoint with guided parameter selection.
+
+**Codex Review Parameters** (from `codex review --help`):
+| Parameter | Description |
+|-----------|-------------|
+| `[PROMPT]` | Custom review instructions (positional) |
+| `-c model=<model>` | Override model via config |
+| `--uncommitted` | Review staged, unstaged, and untracked changes |
+| `--base <BRANCH>` | Review changes against base branch |
+| `--commit <SHA>` | Review changes introduced by a commit |
+| `--title <TITLE>` | Optional commit title for review summary |
+
+## Prompt Template Format
+
+Follow the standard ccw cli prompt template:
+
+```
+PURPOSE: [what] + [why] + [success criteria] + [constraints/scope]
+TASK: • [step 1] • [step 2] • [step 3]
+MODE: review
+CONTEXT: [review target description] | Memory: [relevant context]
+EXPECTED: [deliverable format] + [quality criteria]
+CONSTRAINTS: [focus constraints]
+```
+
+## EXECUTION INSTRUCTIONS - START HERE
+
+**When this command is triggered, follow these exact steps:**
+
+### Step 1: Parse Arguments
+
+Check if user provided arguments directly:
+- `--uncommitted` → Record target = uncommitted
+- `--base <branch>` → Record target = base, branch name
+- `--commit <sha>` → Record target = commit, sha value
+- `--model <model>` → Record model selection
+- `--title <title>` → Record title
+- Remaining text → Use as custom focus/prompt
+
+If no target specified → Continue to Step 2 for interactive selection.
+
+### Step 2: Interactive Parameter Selection
+
+**2.1 Review Target Selection**
+
+```javascript
+AskUserQuestion({
+  questions: [{
+    question: "What do you want to review?",
+    header: "Review Target",
+    options: [
+      { label: "Uncommitted changes (Recommended)", description: "Review staged, unstaged, and untracked changes" },
+      { label: "Compare to branch", description: "Review changes against a base branch (e.g., main)" },
+      { label: "Specific commit", description: "Review changes introduced by a specific commit" }
+    ],
+    multiSelect: false
+  }]
+})
+```
+
+**2.2 Branch/Commit Input (if needed)**
+
+If "Compare to branch" selected:
+```javascript
+AskUserQuestion({
+  questions: [{
+    question: "Which base branch to compare against?",
+    header: "Base Branch",
+    options: [
+      { label: "main", description: "Compare against main branch" },
+      { label: "master", description: "Compare against master branch" },
+      { label: "develop", description: "Compare against develop branch" }
+    ],
+    multiSelect: false
+  }]
+})
+```
+
+If "Specific commit" selected:
+- Run `git log --oneline -10` to show recent commits
+- Ask user to provide commit SHA or select from list
+
+**2.3 Model Selection (Optional)**
+
+```javascript
+AskUserQuestion({
+  questions: [{
+    question: "Which model to use for review?",
+    header: "Model",
+    options: [
+      { label: "Default", description: "Use codex default model (gpt-5.2)" },
+      { label: "o3", description: "OpenAI o3 reasoning model" },
+      { label: "gpt-4.1", description: "GPT-4.1 model" },
+      { label: "o4-mini", description: "OpenAI o4-mini (faster)" }
+    ],
+    multiSelect: false
+  }]
+})
+```
+
+**2.4 Review Focus Selection**
+
+```javascript
+AskUserQuestion({
+  questions: [{
+    question: "What should the review focus on?",
+    header: "Focus Area",
+    options: [
+      { label: "General review (Recommended)", description: "Comprehensive review: correctness, style, bugs, docs" },
+      { label: "Security focus", description: "Security vulnerabilities, input validation, auth issues" },
+      { label: "Performance focus", description: "Performance bottlenecks, complexity, resource usage" },
+      { label: "Code quality", description: "Readability, maintainability, SOLID principles" }
+    ],
+    multiSelect: false
+  }]
+})
+```
+
+### Step 3: Build Prompt and Command
+
+**3.1 Construct Prompt Based on Focus**
+
+**General Review Prompt:**
+```
+PURPOSE: Comprehensive code review to identify issues, improve quality, and ensure best practices; success = actionable feedback with clear priorities
+TASK: • Review code correctness and logic errors • Check coding standards and consistency • Identify potential bugs and edge cases • Evaluate documentation completeness
+MODE: review
+CONTEXT: {target_description} | Memory: Project conventions from CLAUDE.md
+EXPECTED: Structured review report with: severity levels (Critical/High/Medium/Low), file:line references, specific improvement suggestions, priority ranking
+CONSTRAINTS: Focus on actionable feedback
+```
+
+**Security Focus Prompt:**
+```
+PURPOSE: Security-focused code review to identify vulnerabilities and security risks; success = all security issues documented with remediation
+TASK: • Scan for injection vulnerabilities (SQL, XSS, command) • Check authentication and authorization logic • Evaluate input validation and sanitization • Identify sensitive data exposure risks
+MODE: review
+CONTEXT: {target_description} | Memory: Security best practices, OWASP Top 10
+EXPECTED: Security report with: vulnerability classification, CVE references where applicable, remediation code snippets, risk severity matrix
+CONSTRAINTS: Security-first analysis | Flag all potential vulnerabilities
+```
+
+**Performance Focus Prompt:**
+```
+PURPOSE: Performance-focused code review to identify bottlenecks and optimization opportunities; success = measurable improvement recommendations
+TASK: • Analyze algorithmic complexity (Big-O) • Identify memory allocation issues • Check for N+1 queries and blocking operations • Evaluate caching opportunities
+MODE: review
+CONTEXT: {target_description} | Memory: Performance patterns and anti-patterns
+EXPECTED: Performance report with: complexity analysis, bottleneck identification, optimization suggestions with expected impact, benchmark recommendations
+CONSTRAINTS: Performance optimization focus
+```
+
+**Code Quality Focus Prompt:**
+```
+PURPOSE: Code quality review to improve maintainability and readability; success = cleaner, more maintainable code
+TASK: • Assess SOLID principles adherence • Identify code duplication and abstraction opportunities • Review naming conventions and clarity • Evaluate test coverage implications
+MODE: review
+CONTEXT: {target_description} | Memory: Project coding standards
+EXPECTED: Quality report with: principle violations, refactoring suggestions, naming improvements, maintainability score
+CONSTRAINTS: Code quality and maintainability focus
+```
+
+**3.2 Build Target Description**
+
+Based on selection, set `{target_description}`:
+- Uncommitted: `Reviewing uncommitted changes (staged + unstaged + untracked)`
+- Base branch: `Reviewing changes against {branch} branch`
+- Commit: `Reviewing changes introduced by commit {sha}`
+
+### Step 4: Execute via CCW CLI
+
+Build and execute the ccw cli command:
+
+```bash
+# Base structure
+ccw cli -p "<PROMPT>" --tool codex --mode review [OPTIONS]
+```
+
+**Command Construction:**
+
+```bash
+# Variables from user selection
+TARGET_FLAG=""      # --uncommitted | --base <branch> | --commit <sha>
+MODEL_FLAG=""       # --model <model> (if not default)
+TITLE_FLAG=""       # --title "<title>" (if provided)
+
+# Build target flag
+if [ "$target" = "uncommitted" ]; then
+  TARGET_FLAG="--uncommitted"
+elif [ "$target" = "base" ]; then
+  TARGET_FLAG="--base $branch"
+elif [ "$target" = "commit" ]; then
+  TARGET_FLAG="--commit $sha"
+fi
+
+# Build model flag (only if not default)
+if [ "$model" != "default" ] && [ -n "$model" ]; then
+  MODEL_FLAG="--model $model"
+fi
+
+# Build title flag (if provided)
+if [ -n "$title" ]; then
+  TITLE_FLAG="--title \"$title\""
+fi
+
+# Execute
+ccw cli -p "$PROMPT" --tool codex --mode review $TARGET_FLAG $MODEL_FLAG $TITLE_FLAG
+```
+
+**Full Example Commands:**
+
+**Option 1: With custom prompt (reviews uncommitted by default):**
+```bash
+ccw cli -p "
+PURPOSE: Comprehensive code review to identify issues and improve quality; success = actionable feedback with priorities
+TASK: • Review correctness and logic • Check standards compliance • Identify bugs and edge cases • Evaluate documentation
+MODE: review
+CONTEXT: Reviewing uncommitted changes | Memory: Project conventions
+EXPECTED: Structured report with severity levels, file:line refs, improvement suggestions
+CONSTRAINTS: Actionable feedback
+" --tool codex --mode review --rule analysis-review-code-quality
+```
+
+**Option 2: Target flag only (no prompt allowed):**
+```bash
+ccw cli --tool codex --mode review --uncommitted
+```
+
+### Step 5: Execute and Display Results
+
+```bash
+Bash({
+  command: "ccw cli -p \"$PROMPT\" --tool codex --mode review $FLAGS",
+  run_in_background: true
+})
+```
+
+Wait for completion and display formatted results.
+
+## Quick Usage Examples
+
+### Direct Execution (No Interaction)
+
+```bash
+# Review uncommitted changes with default settings
+/cli:codex-review --uncommitted
+
+# Review against main branch
+/cli:codex-review --base main
+
+# Review specific commit
+/cli:codex-review --commit abc123
+
+# Review with custom model
+/cli:codex-review --uncommitted --model o3
+
+# Review with security focus
+/cli:codex-review --uncommitted security
+
+# Full options
+/cli:codex-review --base main --model o3 --title "Auth Feature" security
+```
+
+### Interactive Mode
+
+```bash
+# Start interactive selection (guided flow)
+/cli:codex-review
+```
+
+## Focus Area Mapping
+
+| User Selection | Prompt Focus | Key Checks |
+|----------------|--------------|------------|
+| General review | Comprehensive | Correctness, style, bugs, docs |
+| Security focus | Security-first | Injection, auth, validation, exposure |
+| Performance focus | Optimization | Complexity, memory, queries, caching |
+| Code quality | Maintainability | SOLID, duplication, naming, tests |
+
+## Error Handling
+
+### No Changes to Review
+```
+No changes found for review target. Suggestions:
+- For --uncommitted: Make some code changes first
+- For --base: Ensure branch exists and has diverged
+- For --commit: Verify commit SHA exists
+```
+
+### Invalid Branch
+```bash
+# Show available branches
+git branch -a --list | head -20
+```
+
+### Invalid Commit
+```bash
+# Show recent commits
+git log --oneline -10
+```
+
+## Integration Notes
+
+- Uses `ccw cli --tool codex --mode review` endpoint
+- Model passed via prompt (codex uses `-c model=` internally)
+- Target flags (`--uncommitted`, `--base`, `--commit`) passed through to codex
+- Prompt follows standard ccw cli template format for consistency
+
+## Validation Constraints
+
+**IMPORTANT: Target flags and prompt are mutually exclusive**
+
+The codex CLI has a constraint where target flags (`--uncommitted`, `--base`, `--commit`) cannot be used with a positional `[PROMPT]` argument:
+
+```
+error: the argument '--uncommitted' cannot be used with '[PROMPT]'
+error: the argument '--base <BRANCH>' cannot be used with '[PROMPT]'
+error: the argument '--commit <SHA>' cannot be used with '[PROMPT]'
+```
+
+**Behavior:**
+- When ANY target flag is specified, ccw cli automatically skips template concatenation (systemRules/roles)
+- The review uses codex's default review behavior for the specified target
+- Custom prompts are only supported WITHOUT target flags (reviews uncommitted changes by default)
+
+**Valid combinations:**
+| Command | Result |
+|---------|--------|
+| `codex review "Focus on security"` | ✓ Custom prompt, reviews uncommitted (default) |
+| `codex review --uncommitted` | ✓ No prompt, uses default review |
+| `codex review --base main` | ✓ No prompt, uses default review |
+| `codex review --commit abc123` | ✓ No prompt, uses default review |
+| `codex review --uncommitted "prompt"` | ✗ Invalid - mutually exclusive |
+| `codex review --base main "prompt"` | ✗ Invalid - mutually exclusive |
+| `codex review --commit abc123 "prompt"` | ✗ Invalid - mutually exclusive |
+
+**Examples:**
+```bash
+# ✓ Valid: prompt only (reviews uncommitted by default)
+ccw cli -p "Focus on security" --tool codex --mode review
+
+# ✓ Valid: target flag only (no prompt)
+ccw cli --tool codex --mode review --uncommitted
+ccw cli --tool codex --mode review --base main
+ccw cli --tool codex --mode review --commit abc123
+
+# ✗ Invalid: target flag with prompt (will fail)
+ccw cli -p "Review this" --tool codex --mode review --uncommitted
+ccw cli -p "Review this" --tool codex --mode review --base main
+ccw cli -p "Review this" --tool codex --mode review --commit abc123
+```

.claude/commands/issue/discover-by-prompt.md

@@ -267,7 +267,7 @@ EXPECTED: JSON exploration plan following exploration-plan-schema.json:
   "estimated_iterations": N,
   "termination_conditions": [...]
 }
-RULES: $(cat ~/.claude/workflows/cli-templates/protocols/analysis-protocol.md) | Use ACE context to inform targets | Focus on actionable plan
+CONSTRAINTS: Use ACE context to inform targets | Focus on actionable plan
 `;
 
 // Step 3: Execute Gemini planning

.claude/commands/issue/plan.md

@@ -131,7 +131,7 @@ TASK: • Analyze issue titles/tags semantically • Identify functional/archite
 MODE: analysis
 CONTEXT: Issue metadata only
 EXPECTED: JSON with groups array, each containing max 4 issue_ids, theme, rationale
-RULES: $(cat ~/.claude/workflows/cli-templates/protocols/analysis-protocol.md) | Each issue in exactly one group | Max 4 issues per group | Balance group sizes
+CONSTRAINTS: Each issue in exactly one group | Max 4 issues per group | Balance group sizes
 
 INPUT:
 ${JSON.stringify(issueSummaries, null, 2)}
@@ -195,12 +195,26 @@ ${issueList}
 
 ### Workflow
 1. Fetch issue details: ccw issue status <id> --json
-2. Load project context files
-3. Explore codebase (ACE semantic search)
-4. Plan solution with tasks (schema: solution-schema.json)
-5. **If github_url exists**: Add final task to comment on GitHub issue
-6. Write solution to: .workflow/issues/solutions/{issue-id}.jsonl
-7. Single solution → auto-bind; Multiple → return for selection
+2. **Analyze failure history** (if issue.feedback exists):
+   - Extract failure details from issue.feedback (type='failure', stage='execute')
+   - Parse error_type, message, task_id, solution_id from content JSON
+   - Identify failure patterns: repeated errors, root causes, blockers
+   - **Constraint**: Avoid repeating failed approaches
+3. Load project context files
+4. Explore codebase (ACE semantic search)
+5. Plan solution with tasks (schema: solution-schema.json)
+   - **If previous solution failed**: Reference failure analysis in solution.approach
+   -  Add explicit verification steps to prevent same failure mode
+6. **If github_url exists**: Add final task to comment on GitHub issue
+7. Write solution to: .workflow/issues/solutions/{issue-id}.jsonl
+8. Single solution → auto-bind; Multiple → return for selection
+
+### Failure-Aware Planning Rules
+- **Extract failure patterns**: Parse issue.feedback where type='failure' and stage='execute'
+- **Identify root causes**: Analyze error_type (test_failure, compilation, timeout, etc.)
+- **Design alternative approach**: Create solution that addresses root cause
+- **Add prevention steps**: Include explicit verification to catch same error earlier
+- **Document lessons**: Reference previous failures in solution.approach
 
 ### Rules
 - Solution ID format: SOL-{issue-id}-{uid} (uid: 4 random alphanumeric chars, e.g., a7x9)

.claude/commands/issue/queue.md

@@ -65,9 +65,13 @@ Queue formation command using **issue-queue-agent** that analyzes all bound solu
 --queues <n>          Number of parallel queues (default: 1)
 --issue <id>          Form queue for specific issue only
 --append <id>         Append issue to active queue (don't create new)
+--force               Skip active queue check, always create new queue
 
 # CLI subcommands (ccw issue queue ...)
 ccw issue queue list                  List all queues with status
+ccw issue queue add <issue-id>        Add issue to queue (interactive if active queue exists)
+ccw issue queue add <issue-id> -f     Add to new queue without prompt (force)
+ccw issue queue merge <src> --queue <target>  Merge source queue into target queue
 ccw issue queue switch <queue-id>     Switch active queue
 ccw issue queue archive               Archive current queue
 ccw issue queue delete <queue-id>     Delete queue from history
@@ -92,7 +96,7 @@ Phase 2-4: Agent-Driven Queue Formation (issue-queue-agent)
    │   ├─ Build dependency DAG from conflicts
    │   ├─ Calculate semantic priority per solution
    │   └─ Assign execution groups (parallel/sequential)
-   └─ Each agent writes: queue JSON + index update
+   └─ Each agent writes: queue JSON + index update (NOT active yet)
 
 Phase 5: Conflict Clarification (if needed)
    ├─ Collect `clarifications` arrays from all agents
@@ -102,7 +106,24 @@ Phase 5: Conflict Clarification (if needed)
 
 Phase 6: Status Update & Summary
    ├─ Update issue statuses to 'queued'
-   └─ Display queue summary (N queues), next step: /issue:execute
+   └─ Display new queue summary (N queues)
+
+Phase 7: Active Queue Check & Decision (REQUIRED)
+   ├─ Read queue index: ccw issue queue list --brief
+   ├─ Get generated queue ID from agent output
+   ├─ If NO active queue exists:
+   │   ├─ Set generated queue as active_queue_id
+   │   ├─ Update index.json
+   │   └─ Display: "Queue created and activated"
+   │
+   └─ If active queue exists with items:
+       ├─ Display both queues to user
+       ├─ Use AskUserQuestion to prompt:
+       │   ├─ "Use new queue (keep existing)" → Set new as active, keep old inactive
+       │   ├─ "Merge: add new items to existing" → Merge new → existing, delete new
+       │   ├─ "Merge: add existing items to new" → Merge existing → new, archive old
+       │   └─ "Cancel" → Delete new queue, keep existing active
+       └─ Execute chosen action
 ```
 
 ## Implementation
@@ -306,6 +327,41 @@ ccw issue update <issue-id> --status queued
 - Show unplanned issues (planned but NOT in queue)
 - Show next step: `/issue:execute`
 
+### Phase 7: Active Queue Check & Decision
+
+**After agent completes Phase 1-6, check for active queue:**
+
+```bash
+ccw issue queue list --brief
+```
+
+**Decision:**
+- If `active_queue_id` is null → `ccw issue queue switch <new-queue-id>` (activate new queue)
+- If active queue exists → Use **AskUserQuestion** to prompt user
+
+**AskUserQuestion:**
+```javascript
+AskUserQuestion({
+  questions: [{
+    question: "Active queue exists. How would you like to proceed?",
+    header: "Queue Action",
+    options: [
+      { label: "Merge into existing queue", description: "Add new items to active queue, delete new queue" },
+      { label: "Use new queue", description: "Switch to new queue, keep existing in history" },
+      { label: "Cancel", description: "Delete new queue, keep existing active" }
+    ],
+    multiSelect: false
+  }]
+})
+```
+
+**Action Commands:**
+
+| User Choice | Commands |
+|-------------|----------|
+| **Merge into existing** | `ccw issue queue merge <new-queue-id> --queue <active-queue-id>` then `ccw issue queue delete <new-queue-id>` |
+| **Use new queue** | `ccw issue queue switch <new-queue-id>` |
+| **Cancel** | `ccw issue queue delete <new-queue-id>` |
 
 ## Storage Structure (Queue History)
 
@@ -360,6 +416,9 @@ ccw issue update <issue-id> --status queued
 | User cancels clarification | Abort queue formation |
 | **index.json not updated** | Auto-fix: Set active_queue_id to new queue |
 | **Queue file missing solutions** | Abort with error, agent must regenerate |
+| **User cancels queue add** | Display message, return without changes |
+| **Merge with empty source** | Skip merge, display warning |
+| **All items duplicate** | Skip merge, display "All items already exist" |
 
 ## Quality Checklist
 

.claude/commands/memory/swagger-docs.md

@@ -223,8 +223,8 @@ TASK:
 MODE: analysis
 CONTEXT: @src/**/*.controller.ts @src/**/*.routes.ts @src/**/*.dto.ts @src/**/middleware/**/*
 EXPECTED: JSON format API structure analysis report with modules, endpoints, security schemes, and error codes
-RULES: $(cat ~/.claude/workflows/cli-templates/protocols/analysis-protocol.md) | Strict RESTful standards | Identify all public endpoints | Document output language: {lang}
-" --tool gemini --mode analysis --cd {project_root}
+CONSTRAINTS: Strict RESTful standards | Identify all public endpoints | Document output language: {lang}
+" --tool gemini --mode analysis --rule analysis-code-patterns --cd {project_root}
 ```
 
 **Update swagger-planning-data.json** with analysis results:
@@ -387,7 +387,7 @@ bash(cat ${session_dir}/.process/swagger-planning-data.json | jq -r '.api_struct
         "step": 1,
         "title": "Generate OpenAPI spec file",
         "description": "Create complete swagger.yaml specification file",
-        "cli_prompt": "PURPOSE: Generate OpenAPI 3.0.3 specification file from analyzed API structure\nTASK:\n• Define openapi version: 3.0.3\n• Define info: title, description, version, contact, license\n• Define servers: development, staging, production environments\n• Define tags: organized by business modules\n• Define paths: all API endpoints with complete specifications\n• Define components: schemas, securitySchemes, parameters, responses\nMODE: write\nCONTEXT: @[api_analysis]\nEXPECTED: Complete swagger.yaml file following OpenAPI 3.0.3 specification\nRULES: $(cat ~/.claude/workflows/cli-templates/protocols/write-protocol.md) $(cat ~/.claude/workflows/cli-templates/prompts/documentation/swagger-api.txt) | Use {lang} for all descriptions | Strict RESTful standards",
+        "cli_prompt": "PURPOSE: Generate OpenAPI 3.0.3 specification file from analyzed API structure\nTASK:\n• Define openapi version: 3.0.3\n• Define info: title, description, version, contact, license\n• Define servers: development, staging, production environments\n• Define tags: organized by business modules\n• Define paths: all API endpoints with complete specifications\n• Define components: schemas, securitySchemes, parameters, responses\nMODE: write\nCONTEXT: @[api_analysis]\nEXPECTED: Complete swagger.yaml file following OpenAPI 3.0.3 specification\nCONSTRAINTS: Use {lang} for all descriptions | Strict RESTful standards\n--rule documentation-swagger-api",
         "output": "swagger.yaml"
       }
     ],
@@ -429,7 +429,7 @@ bash(cat ${session_dir}/.process/swagger-planning-data.json | jq -r '.api_struct
       {
         "step": 1,
         "title": "Generate authentication documentation",
-        "cli_prompt": "PURPOSE: Generate comprehensive authentication documentation for API security\nTASK:\n• Document authentication mechanism: JWT Bearer Token\n• Explain header format: Authorization: Bearer <token>\n• Describe token lifecycle: acquisition, refresh, expiration handling\n• Define permission levels: public, user, admin, super_admin\n• Document authentication failure responses: 401/403 error handling\nMODE: write\nCONTEXT: @[auth_patterns] @src/**/auth/**/* @src/**/guard/**/*\nEXPECTED: Complete authentication guide in {lang}\nRULES: $(cat ~/.claude/workflows/cli-templates/protocols/write-protocol.md) | Include code examples | Clear step-by-step instructions",
+        "cli_prompt": "PURPOSE: Generate comprehensive authentication documentation for API security\nTASK:\n• Document authentication mechanism: JWT Bearer Token\n• Explain header format: Authorization: Bearer <token>\n• Describe token lifecycle: acquisition, refresh, expiration handling\n• Define permission levels: public, user, admin, super_admin\n• Document authentication failure responses: 401/403 error handling\nMODE: write\nCONTEXT: @[auth_patterns] @src/**/auth/**/* @src/**/guard/**/*\nEXPECTED: Complete authentication guide in {lang}\nCONSTRAINTS: Include code examples | Clear step-by-step instructions\n--rule development-feature",
         "output": "{auth_doc_name}"
       }
     ],
@@ -464,7 +464,7 @@ bash(cat ${session_dir}/.process/swagger-planning-data.json | jq -r '.api_struct
       {
         "step": 1,
         "title": "Generate error code specification document",
-        "cli_prompt": "PURPOSE: Generate comprehensive error code specification for consistent API error handling\nTASK:\n• Define error response format: {code, message, details, timestamp}\n• Document authentication errors (AUTH_xxx): 401/403 series\n• Document parameter errors (PARAM_xxx): 400 series\n• Document business errors (BIZ_xxx): business logic errors\n• Document system errors (SYS_xxx): 500 series\n• For each error code: HTTP status, error message, possible causes, resolution suggestions\nMODE: write\nCONTEXT: @src/**/*.exception.ts @src/**/*.filter.ts\nEXPECTED: Complete error code specification in {lang} with tables and examples\nRULES: $(cat ~/.claude/workflows/cli-templates/protocols/write-protocol.md) | Include response examples | Clear categorization",
+        "cli_prompt": "PURPOSE: Generate comprehensive error code specification for consistent API error handling\nTASK:\n• Define error response format: {code, message, details, timestamp}\n• Document authentication errors (AUTH_xxx): 401/403 series\n• Document parameter errors (PARAM_xxx): 400 series\n• Document business errors (BIZ_xxx): business logic errors\n• Document system errors (SYS_xxx): 500 series\n• For each error code: HTTP status, error message, possible causes, resolution suggestions\nMODE: write\nCONTEXT: @src/**/*.exception.ts @src/**/*.filter.ts\nEXPECTED: Complete error code specification in {lang} with tables and examples\nCONSTRAINTS: Include response examples | Clear categorization\n--rule development-feature",
         "output": "{error_doc_name}"
       }
     ],
@@ -523,7 +523,7 @@ bash(cat ${session_dir}/.process/swagger-planning-data.json | jq -r '.api_struct
         "step": 1,
         "title": "Generate module API documentation",
         "description": "Generate complete API documentation for ${module_name}",
-        "cli_prompt": "PURPOSE: Generate complete RESTful API documentation for ${module_name} module\nTASK:\n• Create module overview: purpose, use cases, prerequisites\n• Generate endpoint index: grouped by functionality\n• For each endpoint document:\n  - Functional description: purpose and business context\n  - Request method: GET/POST/PUT/DELETE\n  - URL path: complete API path\n  - Request headers: Authorization and other required headers\n  - Path parameters: {id} and other path variables\n  - Query parameters: pagination, filters, etc.\n  - Request body: JSON Schema format\n  - Response body: success and error responses\n  - Field description table: type, required, example, description\n• Add usage examples: cURL, JavaScript, Python\n• Add version info: v1.0.0, last updated date\nMODE: write\nCONTEXT: @[module_endpoints] @[source_code]\nEXPECTED: Complete module API documentation in {lang} with all endpoints fully documented\nRULES: $(cat ~/.claude/workflows/cli-templates/protocols/write-protocol.md) $(cat ~/.claude/workflows/cli-templates/prompts/documentation/swagger-api.txt) | RESTful standards | Include all response codes",
+        "cli_prompt": "PURPOSE: Generate complete RESTful API documentation for ${module_name} module\nTASK:\n• Create module overview: purpose, use cases, prerequisites\n• Generate endpoint index: grouped by functionality\n• For each endpoint document:\n  - Functional description: purpose and business context\n  - Request method: GET/POST/PUT/DELETE\n  - URL path: complete API path\n  - Request headers: Authorization and other required headers\n  - Path parameters: {id} and other path variables\n  - Query parameters: pagination, filters, etc.\n  - Request body: JSON Schema format\n  - Response body: success and error responses\n  - Field description table: type, required, example, description\n• Add usage examples: cURL, JavaScript, Python\n• Add version info: v1.0.0, last updated date\nMODE: write\nCONTEXT: @[module_endpoints] @[source_code]\nEXPECTED: Complete module API documentation in {lang} with all endpoints fully documented\nCONSTRAINTS: RESTful standards | Include all response codes\n--rule documentation-swagger-api",
         "output": "${module_doc_name}"
       }
     ],
@@ -559,7 +559,7 @@ bash(cat ${session_dir}/.process/swagger-planning-data.json | jq -r '.api_struct
       {
         "step": 1,
         "title": "Generate API overview",
-        "cli_prompt": "PURPOSE: Generate API overview document with navigation and quick start guide\nTASK:\n• Create introduction: system features, tech stack, version\n• Write quick start guide: authentication, first request example\n• Build module navigation: categorized links to all modules\n• Document environment configuration: development, staging, production\n• List SDKs and tools: client libraries, Postman collection\nMODE: write\nCONTEXT: @[all_module_docs] @.workflow/docs/${project_name}/api/swagger.yaml\nEXPECTED: Complete API overview in {lang} with navigation links\nRULES: $(cat ~/.claude/workflows/cli-templates/protocols/write-protocol.md) | Clear structure | Quick start focus",
+        "cli_prompt": "PURPOSE: Generate API overview document with navigation and quick start guide\nTASK:\n• Create introduction: system features, tech stack, version\n• Write quick start guide: authentication, first request example\n• Build module navigation: categorized links to all modules\n• Document environment configuration: development, staging, production\n• List SDKs and tools: client libraries, Postman collection\nMODE: write\nCONTEXT: @[all_module_docs] @.workflow/docs/${project_name}/api/swagger.yaml\nEXPECTED: Complete API overview in {lang} with navigation links\nCONSTRAINTS: Clear structure | Quick start focus\n--rule development-feature",
         "output": "README.md"
       }
     ],
@@ -602,7 +602,7 @@ bash(cat ${session_dir}/.process/swagger-planning-data.json | jq -r '.api_struct
       {
         "step": 1,
         "title": "Generate test report",
-        "cli_prompt": "PURPOSE: Generate comprehensive API test validation report\nTASK:\n• Document test environment configuration\n• Calculate endpoint coverage statistics\n• Report test results: pass/fail counts\n• Document boundary tests: parameter limits, null values, special characters\n• Document exception tests: auth failures, permission denied, resource not found\n• List issues found with recommendations\nMODE: write\nCONTEXT: @[swagger_spec]\nEXPECTED: Complete test report in {lang} with detailed results\nRULES: $(cat ~/.claude/workflows/cli-templates/protocols/write-protocol.md) | Include test cases | Clear pass/fail status",
+        "cli_prompt": "PURPOSE: Generate comprehensive API test validation report\nTASK:\n• Document test environment configuration\n• Calculate endpoint coverage statistics\n• Report test results: pass/fail counts\n• Document boundary tests: parameter limits, null values, special characters\n• Document exception tests: auth failures, permission denied, resource not found\n• List issues found with recommendations\nMODE: write\nCONTEXT: @[swagger_spec]\nEXPECTED: Complete test report in {lang} with detailed results\nCONSTRAINTS: Include test cases | Clear pass/fail status\n--rule development-tests",
         "output": "{test_doc_name}"
       }
     ],

.claude/commands/memory/tech-research-rules.md

@@ -147,8 +147,8 @@ You are generating path-conditional rules for Claude Code.
 
 ## Instructions
 
-Read the agent prompt template for detailed instructions:
-$(cat ~/.claude/workflows/cli-templates/prompts/rules/tech-rules-agent-prompt.txt)
+Read the agent prompt template for detailed instructions.
+Use --rule rules-tech-rules-agent-prompt to load the template automatically.
 
 ## Execution Steps
 

.claude/commands/workflow/debug-with-file.md

@@ -0,0 +1,666 @@
+---
+name: debug-with-file
+description: Interactive hypothesis-driven debugging with documented exploration, understanding evolution, and Gemini-assisted correction
+argument-hint: "\"bug description or error message\""
+allowed-tools: TodoWrite(*), Task(*), AskUserQuestion(*), Read(*), Grep(*), Glob(*), Bash(*), Edit(*), Write(*)
+---
+
+# Workflow Debug-With-File Command (/workflow:debug-with-file)
+
+## Overview
+
+Enhanced evidence-based debugging with **documented exploration process**. Records understanding evolution, consolidates insights, and uses Gemini to correct misunderstandings.
+
+**Core workflow**: Explore → Document → Log → Analyze → Correct Understanding → Fix → Verify
+
+**Key enhancements over /workflow:debug**:
+- **understanding.md**: Timeline of exploration and learning
+- **Gemini-assisted correction**: Validates and corrects hypotheses
+- **Consolidation**: Simplifies proven-wrong understanding to avoid clutter
+- **Learning retention**: Preserves what was learned, even from failed attempts
+
+## Usage
+
+```bash
+/workflow:debug-with-file <BUG_DESCRIPTION>
+
+# Arguments
+<bug-description>          Bug description, error message, or stack trace (required)
+```
+
+## Execution Process
+
+```
+Session Detection:
+   ├─ Check if debug session exists for this bug
+   ├─ EXISTS + understanding.md exists → Continue mode
+   └─ NOT_FOUND → Explore mode
+
+Explore Mode:
+   ├─ Locate error source in codebase
+   ├─ Document initial understanding in understanding.md
+   ├─ Generate testable hypotheses with Gemini validation
+   ├─ Add NDJSON logging instrumentation
+   └─ Output: Hypothesis list + await user reproduction
+
+Analyze Mode:
+   ├─ Parse debug.log, validate each hypothesis
+   ├─ Use Gemini to analyze evidence and correct understanding
+   ├─ Update understanding.md with:
+   │   ├─ New evidence
+   │   ├─ Corrected misunderstandings (strikethrough + correction)
+   │   └─ Consolidated current understanding
+   └─ Decision:
+       ├─ Confirmed → Fix root cause
+       ├─ Inconclusive → Add more logging, iterate
+       └─ All rejected → Gemini-assisted new hypotheses
+
+Fix & Cleanup:
+   ├─ Apply fix based on confirmed hypothesis
+   ├─ User verifies
+   ├─ Document final understanding + lessons learned
+   ├─ Remove debug instrumentation
+   └─ If not fixed → Return to Analyze mode
+```
+
+## Implementation
+
+### Session Setup & Mode Detection
+
+```javascript
+const getUtc8ISOString = () => new Date(Date.now() + 8 * 60 * 60 * 1000).toISOString()
+
+const bugSlug = bug_description.toLowerCase().replace(/[^a-z0-9]+/g, '-').substring(0, 30)
+const dateStr = getUtc8ISOString().substring(0, 10)
+
+const sessionId = `DBG-${bugSlug}-${dateStr}`
+const sessionFolder = `.workflow/.debug/${sessionId}`
+const debugLogPath = `${sessionFolder}/debug.log`
+const understandingPath = `${sessionFolder}/understanding.md`
+const hypothesesPath = `${sessionFolder}/hypotheses.json`
+
+// Auto-detect mode
+const sessionExists = fs.existsSync(sessionFolder)
+const hasUnderstanding = sessionExists && fs.existsSync(understandingPath)
+const logHasContent = sessionExists && fs.existsSync(debugLogPath) && fs.statSync(debugLogPath).size > 0
+
+const mode = logHasContent ? 'analyze' : (hasUnderstanding ? 'continue' : 'explore')
+
+if (!sessionExists) {
+  bash(`mkdir -p ${sessionFolder}`)
+}
+```
+
+---
+
+### Explore Mode
+
+**Step 1.1: Locate Error Source**
+
+```javascript
+// Extract keywords from bug description
+const keywords = extractErrorKeywords(bug_description)
+
+// Search codebase for error locations
+const searchResults = []
+for (const keyword of keywords) {
+  const results = Grep({ pattern: keyword, path: ".", output_mode: "content", "-C": 3 })
+  searchResults.push({ keyword, results })
+}
+
+// Identify affected files and functions
+const affectedLocations = analyzeSearchResults(searchResults)
+```
+
+**Step 1.2: Document Initial Understanding**
+
+Create `understanding.md` with exploration timeline:
+
+```markdown
+# Understanding Document
+
+**Session ID**: ${sessionId}
+**Bug Description**: ${bug_description}
+**Started**: ${getUtc8ISOString()}
+
+---
+
+## Exploration Timeline
+
+### Iteration 1 - Initial Exploration (${timestamp})
+
+#### Current Understanding
+
+Based on bug description and initial code search:
+
+- Error pattern: ${errorPattern}
+- Affected areas: ${affectedLocations.map(l => l.file).join(', ')}
+- Initial hypothesis: ${initialThoughts}
+
+#### Evidence from Code Search
+
+${searchResults.map(r => `
+**Keyword: "${r.keyword}"**
+- Found in: ${r.results.files.join(', ')}
+- Key findings: ${r.insights}
+`).join('\n')}
+
+#### Next Steps
+
+- Generate testable hypotheses
+- Add instrumentation
+- Await reproduction
+
+---
+
+## Current Consolidated Understanding
+
+${initialConsolidatedUnderstanding}
+```
+
+**Step 1.3: Gemini-Assisted Hypothesis Generation**
+
+```bash
+ccw cli -p "
+PURPOSE: Generate debugging hypotheses for: ${bug_description}
+Success criteria: Testable hypotheses with clear evidence criteria
+
+TASK:
+• Analyze error pattern and code search results
+• Identify 3-5 most likely root causes
+• For each hypothesis, specify:
+  - What might be wrong
+  - What evidence would confirm/reject it
+  - Where to add instrumentation
+• Rank by likelihood
+
+MODE: analysis
+
+CONTEXT: @${sessionFolder}/understanding.md | Search results in understanding.md
+
+EXPECTED:
+- Structured hypothesis list (JSON format)
+- Each hypothesis with: id, description, testable_condition, logging_point, evidence_criteria
+- Likelihood ranking (1=most likely)
+
+CONSTRAINTS: Focus on testable conditions
+" --tool gemini --mode analysis --rule analysis-diagnose-bug-root-cause
+```
+
+Save Gemini output to `hypotheses.json`:
+
+```json
+{
+  "iteration": 1,
+  "timestamp": "2025-01-21T10:00:00+08:00",
+  "hypotheses": [
+    {
+      "id": "H1",
+      "description": "Data structure mismatch - expected key not present",
+      "testable_condition": "Check if target key exists in dict",
+      "logging_point": "file.py:func:42",
+      "evidence_criteria": {
+        "confirm": "data shows missing key",
+        "reject": "key exists with valid value"
+      },
+      "likelihood": 1,
+      "status": "pending"
+    }
+  ],
+  "gemini_insights": "...",
+  "corrected_assumptions": []
+}
+```
+
+**Step 1.4: Add NDJSON Instrumentation**
+
+For each hypothesis, add logging (same as original debug command).
+
+**Step 1.5: Update understanding.md**
+
+Append hypothesis section:
+
+```markdown
+#### Hypotheses Generated (Gemini-Assisted)
+
+${hypotheses.map(h => `
+**${h.id}** (Likelihood: ${h.likelihood}): ${h.description}
+- Logging at: ${h.logging_point}
+- Testing: ${h.testable_condition}
+- Evidence to confirm: ${h.evidence_criteria.confirm}
+- Evidence to reject: ${h.evidence_criteria.reject}
+`).join('\n')}
+
+**Gemini Insights**: ${geminiInsights}
+```
+
+---
+
+### Analyze Mode
+
+**Step 2.1: Parse Debug Log**
+
+```javascript
+// Parse NDJSON log
+const entries = Read(debugLogPath).split('\n')
+  .filter(l => l.trim())
+  .map(l => JSON.parse(l))
+
+// Group by hypothesis
+const byHypothesis = groupBy(entries, 'hid')
+```
+
+**Step 2.2: Gemini-Assisted Evidence Analysis**
+
+```bash
+ccw cli -p "
+PURPOSE: Analyze debug log evidence to validate/correct hypotheses for: ${bug_description}
+Success criteria: Clear verdict per hypothesis + corrected understanding
+
+TASK:
+• Parse log entries by hypothesis
+• Evaluate evidence against expected criteria
+• Determine verdict: confirmed | rejected | inconclusive
+• Identify incorrect assumptions from previous understanding
+• Suggest corrections to understanding
+
+MODE: analysis
+
+CONTEXT:
+@${debugLogPath}
+@${understandingPath}
+@${hypothesesPath}
+
+EXPECTED:
+- Per-hypothesis verdict with reasoning
+- Evidence summary
+- List of incorrect assumptions with corrections
+- Updated consolidated understanding
+- Root cause if confirmed, or next investigation steps
+
+CONSTRAINTS: Evidence-based reasoning only, no speculation
+" --tool gemini --mode analysis --rule analysis-diagnose-bug-root-cause
+```
+
+**Step 2.3: Update Understanding with Corrections**
+
+Append new iteration to `understanding.md`:
+
+```markdown
+### Iteration ${n} - Evidence Analysis (${timestamp})
+
+#### Log Analysis Results
+
+${results.map(r => `
+**${r.id}**: ${r.verdict.toUpperCase()}
+- Evidence: ${JSON.stringify(r.evidence)}
+- Reasoning: ${r.reason}
+`).join('\n')}
+
+#### Corrected Understanding
+
+Previous misunderstandings identified and corrected:
+
+${corrections.map(c => `
+- ~~${c.wrong}~~ → ${c.corrected}
+  - Why wrong: ${c.reason}
+  - Evidence: ${c.evidence}
+`).join('\n')}
+
+#### New Insights
+
+${newInsights.join('\n- ')}
+
+#### Gemini Analysis
+
+${geminiAnalysis}
+
+${confirmedHypothesis ? `
+#### Root Cause Identified
+
+**${confirmedHypothesis.id}**: ${confirmedHypothesis.description}
+
+Evidence supporting this conclusion:
+${confirmedHypothesis.supportingEvidence}
+` : `
+#### Next Steps
+
+${nextSteps}
+`}
+
+---
+
+## Current Consolidated Understanding (Updated)
+
+${consolidatedUnderstanding}
+```
+
+**Step 2.4: Consolidate Understanding**
+
+At the bottom of `understanding.md`, update the consolidated section:
+
+- Remove or simplify proven-wrong assumptions
+- Keep them in strikethrough for reference
+- Focus on current valid understanding
+- Avoid repeating details from timeline
+
+```markdown
+## Current Consolidated Understanding
+
+### What We Know
+
+- ${validUnderstanding1}
+- ${validUnderstanding2}
+
+### What Was Disproven
+
+- ~~Initial assumption: ${wrongAssumption}~~ (Evidence: ${disproofEvidence})
+
+### Current Investigation Focus
+
+${currentFocus}
+
+### Remaining Questions
+
+- ${openQuestion1}
+- ${openQuestion2}
+```
+
+**Step 2.5: Update hypotheses.json**
+
+```json
+{
+  "iteration": 2,
+  "timestamp": "2025-01-21T10:15:00+08:00",
+  "hypotheses": [
+    {
+      "id": "H1",
+      "status": "rejected",
+      "verdict_reason": "Evidence shows key exists with valid value",
+      "evidence": {...}
+    },
+    {
+      "id": "H2",
+      "status": "confirmed",
+      "verdict_reason": "Log data confirms timing issue",
+      "evidence": {...}
+    }
+  ],
+  "gemini_corrections": [
+    {
+      "wrong_assumption": "...",
+      "corrected_to": "...",
+      "reason": "..."
+    }
+  ]
+}
+```
+
+---
+
+### Fix & Verification
+
+**Step 3.1: Apply Fix**
+
+(Same as original debug command)
+
+**Step 3.2: Document Resolution**
+
+Append to `understanding.md`:
+
+```markdown
+### Iteration ${n} - Resolution (${timestamp})
+
+#### Fix Applied
+
+- Modified files: ${modifiedFiles.join(', ')}
+- Fix description: ${fixDescription}
+- Root cause addressed: ${rootCause}
+
+#### Verification Results
+
+${verificationResults}
+
+#### Lessons Learned
+
+What we learned from this debugging session:
+
+1. ${lesson1}
+2. ${lesson2}
+3. ${lesson3}
+
+#### Key Insights for Future
+
+- ${insight1}
+- ${insight2}
+```
+
+**Step 3.3: Cleanup**
+
+Remove debug instrumentation (same as original command).
+
+---
+
+## Session Folder Structure
+
+```
+.workflow/.debug/DBG-{slug}-{date}/
+├── debug.log           # NDJSON log (execution evidence)
+├── understanding.md    # NEW: Exploration timeline + consolidated understanding
+├── hypotheses.json     # NEW: Hypothesis history with verdicts
+└── resolution.md       # Optional: Final summary
+```
+
+## Understanding Document Template
+
+```markdown
+# Understanding Document
+
+**Session ID**: DBG-xxx-2025-01-21
+**Bug Description**: [original description]
+**Started**: 2025-01-21T10:00:00+08:00
+
+---
+
+## Exploration Timeline
+
+### Iteration 1 - Initial Exploration (2025-01-21 10:00)
+
+#### Current Understanding
+...
+
+#### Evidence from Code Search
+...
+
+#### Hypotheses Generated (Gemini-Assisted)
+...
+
+### Iteration 2 - Evidence Analysis (2025-01-21 10:15)
+
+#### Log Analysis Results
+...
+
+#### Corrected Understanding
+- ~~[wrong]~~ → [corrected]
+
+#### Gemini Analysis
+...
+
+---
+
+## Current Consolidated Understanding
+
+### What We Know
+- [valid understanding points]
+
+### What Was Disproven
+- ~~[disproven assumptions]~~
+
+### Current Investigation Focus
+[current focus]
+
+### Remaining Questions
+- [open questions]
+```
+
+## Iteration Flow
+
+```
+First Call (/workflow:debug-with-file "error"):
+   ├─ No session exists → Explore mode
+   ├─ Extract error keywords, search codebase
+   ├─ Document initial understanding in understanding.md
+   ├─ Use Gemini to generate hypotheses
+   ├─ Add logging instrumentation
+   └─ Await user reproduction
+
+After Reproduction (/workflow:debug-with-file "error"):
+   ├─ Session exists + debug.log has content → Analyze mode
+   ├─ Parse log, use Gemini to evaluate hypotheses
+   ├─ Update understanding.md with:
+   │   ├─ Evidence analysis results
+   │   ├─ Corrected misunderstandings (strikethrough)
+   │   ├─ New insights
+   │   └─ Updated consolidated understanding
+   ├─ Update hypotheses.json with verdicts
+   └─ Decision:
+       ├─ Confirmed → Fix → Document resolution
+       ├─ Inconclusive → Add logging, document next steps
+       └─ All rejected → Gemini-assisted new hypotheses
+
+Output:
+   ├─ .workflow/.debug/DBG-{slug}-{date}/debug.log
+   ├─ .workflow/.debug/DBG-{slug}-{date}/understanding.md (evolving document)
+   └─ .workflow/.debug/DBG-{slug}-{date}/hypotheses.json (history)
+```
+
+## Gemini Integration Points
+
+### 1. Hypothesis Generation (Explore Mode)
+
+**Purpose**: Generate evidence-based, testable hypotheses
+
+**Prompt Pattern**:
+```
+PURPOSE: Generate debugging hypotheses + evidence criteria
+TASK: Analyze error + code → testable hypotheses with clear pass/fail criteria
+CONTEXT: @understanding.md (search results)
+EXPECTED: JSON with hypotheses, likelihood ranking, evidence criteria
+```
+
+### 2. Evidence Analysis (Analyze Mode)
+
+**Purpose**: Validate hypotheses and correct misunderstandings
+
+**Prompt Pattern**:
+```
+PURPOSE: Analyze debug log evidence + correct understanding
+TASK: Evaluate each hypothesis → identify wrong assumptions → suggest corrections
+CONTEXT: @debug.log @understanding.md @hypotheses.json
+EXPECTED: Verdicts + corrections + updated consolidated understanding
+```
+
+### 3. New Hypothesis Generation (After All Rejected)
+
+**Purpose**: Generate new hypotheses based on what was disproven
+
+**Prompt Pattern**:
+```
+PURPOSE: Generate new hypotheses given disproven assumptions
+TASK: Review rejected hypotheses → identify knowledge gaps → new investigation angles
+CONTEXT: @understanding.md (with disproven section) @hypotheses.json
+EXPECTED: New hypotheses avoiding previously rejected paths
+```
+
+## Error Correction Mechanism
+
+### Correction Format in understanding.md
+
+```markdown
+#### Corrected Understanding
+
+- ~~Assumed dict key "config" was missing~~ → Key exists, but value is None
+  - Why wrong: Only checked existence, not value validity
+  - Evidence: H1 log shows {"config": null, "exists": true}
+
+- ~~Thought error occurred in initialization~~ → Error happens during runtime update
+  - Why wrong: Stack trace misread as init code
+  - Evidence: H2 timestamp shows 30s after startup
+```
+
+### Consolidation Rules
+
+When updating "Current Consolidated Understanding":
+
+1. **Simplify disproven items**: Move to "What Was Disproven" with single-line summary
+2. **Keep valid insights**: Promote confirmed findings to "What We Know"
+3. **Avoid duplication**: Don't repeat timeline details in consolidated section
+4. **Focus on current state**: What do we know NOW, not the journey
+5. **Preserve key corrections**: Keep important wrong→right transformations for learning
+
+**Bad (cluttered)**:
+```markdown
+## Current Consolidated Understanding
+
+In iteration 1 we thought X, but in iteration 2 we found Y, then in iteration 3...
+Also we checked A and found B, and then we checked C...
+```
+
+**Good (consolidated)**:
+```markdown
+## Current Consolidated Understanding
+
+### What We Know
+- Error occurs during runtime update, not initialization
+- Config value is None (not missing key)
+
+### What Was Disproven
+- ~~Initialization error~~ (Timing evidence)
+- ~~Missing key hypothesis~~ (Key exists)
+
+### Current Investigation Focus
+Why is config value None during update?
+```
+
+## Post-Completion Expansion
+
+完成后询问用户是否扩展为issue(test/enhance/refactor/doc)，选中项调用 `/issue:new "{summary} - {dimension}"`
+
+---
+
+## Error Handling
+
+| Situation | Action |
+|-----------|--------|
+| Empty debug.log | Verify reproduction triggered the code path |
+| All hypotheses rejected | Use Gemini to generate new hypotheses based on disproven assumptions |
+| Fix doesn't work | Document failed fix attempt, iterate with refined understanding |
+| >5 iterations | Review consolidated understanding, escalate to `/workflow:lite-fix` with full context |
+| Gemini unavailable | Fallback to manual hypothesis generation, document without Gemini insights |
+| Understanding too long | Consolidate aggressively, archive old iterations to separate file |
+
+## Comparison with /workflow:debug
+
+| Feature | /workflow:debug | /workflow:debug-with-file |
+|---------|-----------------|---------------------------|
+| NDJSON logging | ✅ | ✅ |
+| Hypothesis generation | Manual | Gemini-assisted |
+| Exploration documentation | ❌ | ✅ understanding.md |
+| Understanding evolution | ❌ | ✅ Timeline + corrections |
+| Error correction | ❌ | ✅ Strikethrough + reasoning |
+| Consolidated learning | ❌ | ✅ Current understanding section |
+| Hypothesis history | ❌ | ✅ hypotheses.json |
+| Gemini validation | ❌ | ✅ At key decision points |
+
+## Usage Recommendations
+
+Use `/workflow:debug-with-file` when:
+- Complex bugs requiring multiple investigation rounds
+- Learning from debugging process is valuable
+- Team needs to understand debugging rationale
+- Bug might recur, documentation helps prevention
+
+Use `/workflow:debug` when:
+- Simple, quick bugs
+- One-off issues
+- Documentation overhead not needed

.claude/commands/workflow/lite-execute.md

@@ -81,6 +81,7 @@ AskUserQuestion({
       options: [
         { label: "Skip", description: "No review" },
         { label: "Gemini Review", description: "Gemini CLI tool" },
+        { label: "Codex Review", description: "Git-aware review (prompt OR --uncommitted)" },
         { label: "Agent Review", description: "Current agent review" }
       ]
     }
@@ -171,10 +172,23 @@ Output:
 **Operations**:
 - Initialize result tracking for multi-execution scenarios
 - Set up `previousExecutionResults` array for context continuity
+- **In-Memory Mode**: Echo execution strategy from lite-plan for transparency
 
 ```javascript
 // Initialize result tracking
 previousExecutionResults = []
+
+// In-Memory Mode: Echo execution strategy (transparency before execution)
+if (executionContext) {
+  console.log(`
+📋 Execution Strategy (from lite-plan):
+   Method: ${executionContext.executionMethod}
+   Review: ${executionContext.codeReviewTool}
+   Tasks: ${executionContext.planObject.tasks.length}
+   Complexity: ${executionContext.planObject.complexity}
+${executionContext.executorAssignments ? `   Assignments: ${JSON.stringify(executionContext.executorAssignments)}` : ''}
+  `)
+}
 ```
 
 ### Step 2: Task Grouping & Batch Creation
@@ -313,7 +327,7 @@ for (const call of sequential) {
 
 ```javascript
 function buildExecutionPrompt(batch) {
-  // Task template (4 parts: Modification Points → How → Reference → Done)
+  // Task template (6 parts: Modification Points → Why → How → Reference → Risks → Done)
   const formatTask = (t) => `
 ## ${t.title}
 
@@ -322,18 +336,38 @@ function buildExecutionPrompt(batch) {
 ### Modification Points
 ${t.modification_points.map(p => `- **${p.file}** → \`${p.target}\`: ${p.change}`).join('\n')}
 
+${t.rationale ? `
+### Why this approach (Medium/High)
+${t.rationale.chosen_approach}
+${t.rationale.decision_factors?.length > 0 ? `\nKey factors: ${t.rationale.decision_factors.join(', ')}` : ''}
+${t.rationale.tradeoffs ? `\nTradeoffs: ${t.rationale.tradeoffs}` : ''}
+` : ''}
+
 ### How to do it
 ${t.description}
 
 ${t.implementation.map(step => `- ${step}`).join('\n')}
 
+${t.code_skeleton ? `
+### Code skeleton (High)
+${t.code_skeleton.interfaces?.length > 0 ? `**Interfaces**: ${t.code_skeleton.interfaces.map(i => `\`${i.name}\` - ${i.purpose}`).join(', ')}` : ''}
+${t.code_skeleton.key_functions?.length > 0 ? `\n**Functions**: ${t.code_skeleton.key_functions.map(f => `\`${f.signature}\` - ${f.purpose}`).join(', ')}` : ''}
+${t.code_skeleton.classes?.length > 0 ? `\n**Classes**: ${t.code_skeleton.classes.map(c => `\`${c.name}\` - ${c.purpose}`).join(', ')}` : ''}
+` : ''}
+
 ### Reference
 - Pattern: ${t.reference?.pattern || 'N/A'}
 - Files: ${t.reference?.files?.join(', ') || 'N/A'}
 ${t.reference?.examples ? `- Notes: ${t.reference.examples}` : ''}
 
+${t.risks?.length > 0 ? `
+### Risk mitigations (High)
+${t.risks.map(r => `- ${r.description} → **${r.mitigation}**`).join('\n')}
+` : ''}
+
 ### Done when
-${t.acceptance.map(c => `- [ ] ${c}`).join('\n')}`
+${t.acceptance.map(c => `- [ ] ${c}`).join('\n')}
+${t.verification?.success_metrics?.length > 0 ? `\n**Success metrics**: ${t.verification.success_metrics.join(', ')}` : ''}`
 
   // Build prompt
   const sections = []
@@ -350,9 +384,14 @@ ${t.acceptance.map(c => `- [ ] ${c}`).join('\n')}`
   if (clarificationContext) {
     context.push(`### Clarifications\n${Object.entries(clarificationContext).map(([q, a]) => `- ${q}: ${a}`).join('\n')}`)
   }
+  if (executionContext?.planObject?.data_flow?.diagram) {
+    context.push(`### Data Flow\n${executionContext.planObject.data_flow.diagram}`)
+  }
   if (executionContext?.session?.artifacts?.plan) {
     context.push(`### Artifacts\nPlan: ${executionContext.session.artifacts.plan}`)
   }
+  // Project guidelines (user-defined constraints from /workflow:session:solidify)
+  context.push(`### Project Guidelines\n@.workflow/project-guidelines.json`)
   if (context.length > 0) sections.push(`## Context\n${context.join('\n\n')}`)
 
   sections.push(`Complete each task according to its "Done when" checklist.`)
@@ -392,16 +431,8 @@ ccw cli -p "${buildExecutionPrompt(batch)}" --tool codex --mode write
 
 **Execution with fixed IDs** (predictable ID pattern):
 ```javascript
-// Launch CLI in foreground (NOT background)
-// Timeout based on complexity: Low=40min, Medium=60min, High=100min
-const timeoutByComplexity = {
-  "Low": 2400000,    // 40 minutes
-  "Medium": 3600000, // 60 minutes
-  "High": 6000000    // 100 minutes
-}
-
+// Launch CLI in background, wait for task hook callback
 // Generate fixed execution ID: ${sessionId}-${groupId}
-// This enables predictable ID lookup without relying on resume context chains
 const sessionId = executionContext?.session?.id || 'standalone'
 const fixedExecutionId = `${sessionId}-${batch.groupId}`  // e.g., "implement-auth-2025-12-13-P1"
 
@@ -413,16 +444,12 @@ const cli_command = previousCliId
   ? `ccw cli -p "${buildExecutionPrompt(batch)}" --tool codex --mode write --id ${fixedExecutionId} --resume ${previousCliId}`
   : `ccw cli -p "${buildExecutionPrompt(batch)}" --tool codex --mode write --id ${fixedExecutionId}`
 
-bash_result = Bash(
+// Execute in background, stop output and wait for task hook callback
+Bash(
   command=cli_command,
-  timeout=timeoutByComplexity[planObject.complexity] || 3600000
+  run_in_background=true
 )
-
-// Execution ID is now predictable: ${fixedExecutionId}
-// Can also extract from output: "ID: implement-auth-2025-12-13-P1"
-const cliExecutionId = fixedExecutionId
-
-// Update TodoWrite when execution completes
+// STOP HERE - CLI executes in background, task hook will notify on completion
 ```
 
 **Resume on Failure** (with fixed ID):
@@ -460,32 +487,41 @@ Progress tracked at batch level (not individual task level). Icons: ⚡ (paralle
 
 **Skip Condition**: Only run if `codeReviewTool ≠ "Skip"`
 
-**Review Focus**: Verify implementation against plan acceptance criteria
-- Read plan.json for task acceptance criteria
+**Review Focus**: Verify implementation against plan acceptance criteria and verification requirements
+- Read plan.json for task acceptance criteria and verification checklist
 - Check each acceptance criterion is fulfilled
+- Verify success metrics from verification field (Medium/High complexity)
+- Run unit/integration tests specified in verification field
 - Validate code quality and identify issues
-- Ensure alignment with planned approach
+- Ensure alignment with planned approach and risk mitigations
 
 **Operations**:
 - Agent Review: Current agent performs direct review
 - Gemini Review: Execute gemini CLI with review prompt
-- Custom tool: Execute specified CLI tool (qwen, codex, etc.)
+- Codex Review: Two options - (A) with prompt for complex reviews, (B) `--uncommitted` flag only for quick reviews
+- Custom tool: Execute specified CLI tool (qwen, etc.)
 
 **Unified Review Template** (All tools use same standard):
 
 **Review Criteria**:
 - **Acceptance Criteria**: Verify each criterion from plan.tasks[].acceptance
+- **Verification Checklist** (Medium/High): Check unit_tests, integration_tests, success_metrics from plan.tasks[].verification
 - **Code Quality**: Analyze quality, identify issues, suggest improvements
-- **Plan Alignment**: Validate implementation matches planned approach
+- **Plan Alignment**: Validate implementation matches planned approach and risk mitigations
 
 **Shared Prompt Template** (used by all CLI tools):
 ```
-PURPOSE: Code review for implemented changes against plan acceptance criteria
-TASK: • Verify plan acceptance criteria fulfillment • Analyze code quality • Identify issues • Suggest improvements • Validate plan adherence
+PURPOSE: Code review for implemented changes against plan acceptance criteria and verification requirements
+TASK: • Verify plan acceptance criteria fulfillment • Check verification requirements (unit tests, success metrics) • Analyze code quality • Identify issues • Suggest improvements • Validate plan adherence and risk mitigations
 MODE: analysis
-CONTEXT: @**/* @{plan.json} [@{exploration.json}] | Memory: Review lite-execute changes against plan requirements
-EXPECTED: Quality assessment with acceptance criteria verification, issue identification, and recommendations. Explicitly check each acceptance criterion from plan.json tasks.
-RULES: $(cat ~/.claude/workflows/cli-templates/prompts/analysis/02-review-code-quality.txt) | Focus on plan acceptance criteria and plan adherence | analysis=READ-ONLY
+CONTEXT: @**/* @{plan.json} [@{exploration.json}] | Memory: Review lite-execute changes against plan requirements including verification checklist
+EXPECTED: Quality assessment with:
+  - Acceptance criteria verification (all tasks)
+  - Verification checklist validation (Medium/High: unit_tests, integration_tests, success_metrics)
+  - Issue identification
+  - Recommendations
+  Explicitly check each acceptance criterion and verification item from plan.json tasks.
+CONSTRAINTS: Focus on plan acceptance criteria, verification requirements, and plan adherence | analysis=READ-ONLY
 ```
 
 **Tool-Specific Execution** (Apply shared prompt template above):
@@ -504,8 +540,17 @@ ccw cli -p "[Shared Prompt Template with artifacts]" --tool gemini --mode analys
 ccw cli -p "[Shared Prompt Template with artifacts]" --tool qwen --mode analysis
 # Same prompt as Gemini, different execution engine
 
-# Method 4: Codex Review (autonomous)
-ccw cli -p "[Verify plan acceptance criteria at ${plan.json}]" --tool codex --mode write
+# Method 4: Codex Review (git-aware) - Two mutually exclusive options:
+
+# Option A: With custom prompt (reviews uncommitted by default)
+ccw cli -p "[Shared Prompt Template with artifacts]" --tool codex --mode review
+# Use for complex reviews with specific focus areas
+
+# Option B: Target flag only (no prompt allowed)
+ccw cli --tool codex --mode review --uncommitted
+# Quick review of uncommitted changes without custom instructions
+
+# ⚠️ IMPORTANT: -p prompt and target flags (--uncommitted/--base/--commit) are MUTUALLY EXCLUSIVE
 ```
 
 **Multi-Round Review with Fixed IDs**:

.claude/commands/workflow/lite-fix.md

@@ -380,6 +380,7 @@ if (uniqueClarifications.length > 0) {
 const schema = Bash(`cat ~/.claude/workflows/cli-templates/schemas/fix-plan-json-schema.json`)
 
 // Step 2: Generate fix-plan following schema (Claude directly, no agent)
+// For Medium complexity: include rationale + verification (optional, but recommended)
 const fixPlan = {
   summary: "...",
   root_cause: "...",
@@ -389,13 +390,67 @@ const fixPlan = {
   recommended_execution: "Agent",
   severity: severity,
   risk_level: "...",
-  _metadata: { timestamp: getUtc8ISOString(), source: "direct-planning", planning_mode: "direct" }
+
+  // Medium complexity fields (optional for direct planning, auto-filled for Low)
+  ...(severity === "Medium" ? {
+    design_decisions: [
+      {
+        decision: "Use immediate_patch strategy for minimal risk",
+        rationale: "Keeps changes localized and quick to review",
+        tradeoff: "Defers comprehensive refactoring"
+      }
+    ],
+    tasks_with_rationale: {
+      // Each task gets rationale if Medium
+      task_rationale_example: {
+        rationale: {
+          chosen_approach: "Direct fix approach",
+          alternatives_considered: ["Workaround", "Refactor"],
+          decision_factors: ["Minimal impact", "Quick turnaround"],
+          tradeoffs: "Doesn't address underlying issue"
+        },
+        verification: {
+          unit_tests: ["test_bug_fix_basic"],
+          integration_tests: [],
+          manual_checks: ["Reproduce issue", "Verify fix"],
+          success_metrics: ["Issue resolved", "No regressions"]
+        }
+      }
+    }
+  } : {}),
+
+  _metadata: {
+    timestamp: getUtc8ISOString(),
+    source: "direct-planning",
+    planning_mode: "direct",
+    complexity: severity === "Medium" ? "Medium" : "Low"
+  }
 }
 
-// Step 3: Write fix-plan to session folder
+// Step 3: Merge task rationale into tasks array
+if (severity === "Medium") {
+  fixPlan.tasks = fixPlan.tasks.map(task => ({
+    ...task,
+    rationale: fixPlan.tasks_with_rationale[task.id]?.rationale || {
+      chosen_approach: "Standard fix",
+      alternatives_considered: [],
+      decision_factors: ["Correctness", "Simplicity"],
+      tradeoffs: "None"
+    },
+    verification: fixPlan.tasks_with_rationale[task.id]?.verification || {
+      unit_tests: [`test_${task.id}_basic`],
+      integration_tests: [],
+      manual_checks: ["Verify fix works"],
+      success_metrics: ["Test pass"]
+    }
+  }))
+  delete fixPlan.tasks_with_rationale  // Clean up temp field
+}
+
+// Step 4: Write fix-plan to session folder
 Write(`${sessionFolder}/fix-plan.json`, JSON.stringify(fixPlan, null, 2))
 
-// Step 4: MUST continue to Phase 4 (Confirmation) - DO NOT execute code here
+// Step 5: MUST continue to Phase 4 (Confirmation) - DO NOT execute code here
 ```
 
 **High/Critical Severity** - Invoke cli-lite-planning-agent:
@@ -451,11 +506,41 @@ Generate fix-plan.json with:
   - description
   - modification_points: ALL files to modify for this fix (group related changes)
   - implementation (2-5 steps covering all modification_points)
-  - verification (test criteria)
+  - acceptance: Quantified acceptance criteria
   - depends_on: task IDs this task depends on (use sparingly)
+
+  **High/Critical complexity fields per task** (REQUIRED):
+  - rationale:
+    - chosen_approach: Why this fix approach (not alternatives)
+    - alternatives_considered: Other approaches evaluated
+    - decision_factors: Key factors influencing choice
+    - tradeoffs: Known tradeoffs of this approach
+  - verification:
+    - unit_tests: Test names to add/verify
+    - integration_tests: Integration test names
+    - manual_checks: Manual verification steps
+    - success_metrics: Quantified success criteria
+  - risks:
+    - description: Risk description
+    - probability: Low|Medium|High
+    - impact: Low|Medium|High
+    - mitigation: How to mitigate
+    - fallback: Fallback if fix fails
+  - code_skeleton (optional): Key interfaces/functions to implement
+    - interfaces: [{name, definition, purpose}]
+    - key_functions: [{signature, purpose, returns}]
+
+**Top-level High/Critical fields** (REQUIRED):
+- data_flow: How data flows through affected code
+  - diagram: "A → B → C" style flow
+  - stages: [{stage, input, output, component}]
+- design_decisions: Global fix decisions
+  - [{decision, rationale, tradeoff}]
+
 - estimated_time, recommended_execution, severity, risk_level
 - _metadata:
   - timestamp, source, planning_mode
+  - complexity: "High" | "Critical"
   - diagnosis_angles: ${JSON.stringify(manifest.diagnoses.map(d => d.angle))}
 
 ## Task Grouping Rules
@@ -467,11 +552,21 @@ Generate fix-plan.json with:
 
 ## Execution
 1. Read ALL diagnosis files for comprehensive context
-2. Execute CLI planning using Gemini (Qwen fallback)
+2. Execute CLI planning using Gemini (Qwen fallback) with --rule planning-fix-strategy template
 3. Synthesize findings from multiple diagnosis angles
-4. Parse output and structure fix-plan
-5. Write JSON: Write('${sessionFolder}/fix-plan.json', jsonContent)
-6. Return brief completion summary
+4. Generate fix-plan with:
+   - For High/Critical: REQUIRED new fields (rationale, verification, risks, code_skeleton, data_flow, design_decisions)
+   - Each task MUST have rationale (why this fix), verification (how to verify success), and risks (potential issues)
+5. Parse output and structure fix-plan
+6. Write JSON: Write('${sessionFolder}/fix-plan.json', jsonContent)
+7. Return brief completion summary
+
+## Output Format for CLI
+Include these sections in your fix-plan output:
+- Summary, Root Cause, Strategy (existing)
+- Data Flow: Diagram showing affected code paths
+- Design Decisions: Key architectural choices in the fix
+- Tasks: Each with rationale (Medium/High), verification (Medium/High), risks (High), code_skeleton (High)
 `
 )
 ```
@@ -565,7 +660,11 @@ const fixPlan = JSON.parse(Read(`${sessionFolder}/fix-plan.json`))
 executionContext = {
   mode: "bugfix",
   severity: fixPlan.severity,
-  planObject: fixPlan,
+  planObject: {
+    ...fixPlan,
+    // Ensure complexity is set based on severity for new field consumption
+    complexity: fixPlan.complexity || (fixPlan.severity === 'Critical' ? 'High' : (fixPlan.severity === 'High' ? 'High' : 'Medium'))
+  },
   diagnosisContext: diagnoses,
   diagnosisAngles: manifest.diagnoses.map(d => d.angle),
   diagnosisManifest: manifest,

.claude/commands/workflow/lite-lite-lite.md

@@ -1,8 +1,8 @@
 ---
 name: workflow:lite-lite-lite
-description: Ultra-lightweight multi-tool analysis and direct execution. No artifacts, auto tool selection based on task analysis, user-driven iteration via AskUser.
+description: Ultra-lightweight multi-tool analysis and direct execution. No artifacts for simple tasks; auto-creates planning docs in .workflow/.scratchpad/ for complex tasks. Auto tool selection based on task analysis, user-driven iteration via AskUser.
 argument-hint: "<task description>"
-allowed-tools: TodoWrite(*), Task(*), AskUserQuestion(*), Read(*), Bash(*), mcp__ace-tool__search_context(*)
+allowed-tools: TodoWrite(*), Task(*), AskUserQuestion(*), Read(*), Bash(*), Write(*), mcp__ace-tool__search_context(*), mcp__ccw-tools__write_file(*)
 ---
 
 # Ultra-Lite Multi-Tool Workflow
@@ -14,22 +14,23 @@ allowed-tools: TodoWrite(*), Task(*), AskUserQuestion(*), Read(*), Bash(*), mcp_
 /workflow:lite-lite-lite "Refactor payment module for multi-gateway support"
 ```
 
-**Core Philosophy**: Minimal friction, maximum velocity. No files, no artifacts - just analyze and execute.
+**Core Philosophy**: Minimal friction, maximum velocity. Simple tasks = no artifacts. Complex tasks = lightweight planning doc in `.workflow/.scratchpad/`.
 
 ## Overview
 
-**Zero-artifact workflow**: Clarify → Select Tools → Multi-Mode Analysis → Decision → Direct Execution
+**Complexity-aware workflow**: Clarify → Assess Complexity → Select Tools → Multi-Mode Analysis → Decision → Direct Execution
 
-**vs multi-cli-plan**: No IMPL_PLAN.md, plan.json, synthesis.json - all state in memory.
+**vs multi-cli-plan**: No IMPL_PLAN.md, plan.json, synthesis.json - state in memory or lightweight scratchpad doc for complex tasks.
 
 ## Execution Flow
 
 ```
 Phase 1: Clarify Requirements → AskUser for missing details
+Phase 1.5: Assess Complexity → Determine if planning doc needed
 Phase 2: Select Tools (CLI → Mode → Agent) → 3-step selection
 Phase 3: Multi-Mode Analysis → Execute with --resume chaining
 Phase 4: User Decision → Execute / Refine / Change / Cancel
-Phase 5: Direct Execution → No plan files, immediate implementation
+Phase 5: Direct Execution → No plan files (simple) or scratchpad doc (complex)
 ```
 
 ## Phase 1: Clarify Requirements
@@ -58,6 +59,25 @@ mcp__ace-tool__search_context({
 })
 ```
 
+## Phase 1.5: Assess Complexity
+
+| Level | Creates Plan Doc | Trigger Keywords |
+|-------|------------------|------------------|
+| **simple** | ❌ | (default) |
+| **moderate** | ✅ | module, system, service, integration, multiple |
+| **complex** | ✅ | refactor, migrate, security, auth, payment, database |
+
+```javascript
+// Complexity detection (after ACE query)
+const isComplex = /refactor|migrate|security|auth|payment|database/i.test(taskDescription)
+const isModerate = /module|system|service|integration|multiple/i.test(taskDescription) || aceContext?.relevant_files?.length > 2
+
+if (isComplex || isModerate) {
+  const planPath = `.workflow/.scratchpad/lite3-${taskSlug}-${dateStr}.md`
+  // Create planning doc with: Task, Status, Complexity, Analysis Summary, Execution Plan, Progress Log
+}
+```
+
 ## Phase 2: Select Tools
 
 ### Tool Definitions
@@ -167,7 +187,7 @@ TASK: ${tasks.map(t => `• ${t}`).join(' ')}
 MODE: analysis
 CONTEXT: @**/*
 EXPECTED: ${expected}
-RULES: $(cat ~/.claude/workflows/cli-templates/protocols/analysis-protocol.md) | ${rules}
+CONSTRAINTS: ${rules}
 `
 }
 
@@ -308,6 +328,8 @@ function aggregateResults(mode, results) {
         critiques: parseCritiques(results.find(r => r.phase === 'challenge')?.result), riskScore: calculateRiskScore(results) }
   }
 }
+
+// If planPath exists: update Analysis Summary & Execution Plan sections
 ```
 
 ## Phase 4: User Decision
@@ -348,13 +370,14 @@ AskUserQuestion({
     multiSelect: false
   }]
 })
+// If planPath exists: record decision to Decisions Made table
 // Routing: Execute → Phase 5 | Refine → Phase 3 | Change → Phase 2 | Cancel → End
 ```
 
 ## Phase 5: Direct Execution
 
 ```javascript
-// No IMPL_PLAN.md, no plan.json - direct implementation
+// Simple tasks: No artifacts | Complex tasks: Update scratchpad doc
 const executionAgents = agents.filter(a => a.canExecute)
 const executionTool = selectedAgent.canExecute ? selectedAgent : selectedCLIs[0]
 
@@ -373,11 +396,12 @@ TASK: ${extractedTasks.join(' • ')}
 MODE: write
 CONTEXT: @${affectedFiles.join(' @')}
 EXPECTED: Working implementation with all changes applied
-RULES: $(cat ~/.claude/workflows/cli-templates/protocols/write-protocol.md)
+CONSTRAINTS: Follow existing patterns
 " --tool ${executionTool.name} --mode write`,
     run_in_background: false
   })
 }
+// If planPath exists: update Status to completed/failed, append to Progress Log
 ```
 
 ## TodoWrite Structure
@@ -385,6 +409,7 @@ RULES: $(cat ~/.claude/workflows/cli-templates/protocols/write-protocol.md)
 ```javascript
 TodoWrite({ todos: [
   { content: "Phase 1: Clarify requirements", status: "in_progress", activeForm: "Clarifying requirements" },
+  { content: "Phase 1.5: Assess complexity", status: "pending", activeForm: "Assessing complexity" },
   { content: "Phase 2: Select tools", status: "pending", activeForm: "Selecting tools" },
   { content: "Phase 3: Multi-mode analysis", status: "pending", activeForm: "Running analysis" },
   { content: "Phase 4: User decision", status: "pending", activeForm: "Awaiting decision" },
@@ -409,16 +434,19 @@ TodoWrite({ todos: [
 | Task unclear | Default to first CLI + code-developer |
 | Ambiguous task | Force clarification via AskUser |
 | Execution fails | Present error, ask user for direction |
+| Plan doc write fails | Continue without doc (degrade to zero-artifact mode) |
+| Scratchpad dir missing | Auto-create `.workflow/.scratchpad/` |
 
 ## Comparison with multi-cli-plan
 
 | Aspect | lite-lite-lite | multi-cli-plan |
 |--------|----------------|----------------|
-| **Artifacts** | None | IMPL_PLAN.md, plan.json, synthesis.json |
+| **Artifacts** | Conditional (scratchpad doc for complex tasks) | Always (IMPL_PLAN.md, plan.json, synthesis.json) |
 | **Session** | Stateless (--resume chaining) | Persistent session folder |
 | **Tool Selection** | 3-step (CLI → Mode → Agent) | Config-driven fixed tools |
 | **Analysis Modes** | 5 modes with --resume | Fixed synthesis rounds |
-| **Best For** | Quick analysis, adversarial validation | Complex multi-step implementations |
+| **Complexity** | Auto-detected (simple/moderate/complex) | Assumed complex |
+| **Best For** | Quick analysis, simple-to-moderate tasks | Complex multi-step implementations |
 
 ## Post-Completion Expansion
 

.claude/commands/workflow/lite-plan.md

@@ -497,6 +497,7 @@ ${plan.tasks.map((t, i) => `${i+1}. ${t.title} (${t.file})`).join('\n')}
 
 **Step 4.2: Collect Confirmation**
 ```javascript
+// Note: Execution "Other" option allows specifying CLI tools from ~/.claude/cli-tools.json
 AskUserQuestion({
   questions: [
     {
@@ -524,8 +525,9 @@ AskUserQuestion({
       header: "Review",
       multiSelect: false,
       options: [
-        { label: "Gemini Review", description: "Gemini CLI" },
-        { label: "Agent Review", description: "@code-reviewer" },
+        { label: "Gemini Review", description: "Gemini CLI review" },
+        { label: "Codex Review", description: "Git-aware review (prompt OR --uncommitted)" },
+        { label: "Agent Review", description: "@code-reviewer agent" },
         { label: "Skip", description: "No review" }
       ]
     }

.claude/commands/workflow/multi-cli-plan.md

@@ -17,15 +17,13 @@ allowed-tools: TodoWrite(*), Task(*), AskUserQuestion(*), Read(*), Bash(*), Writ
 /workflow:multi-cli-plan "Add dark mode support" --max-rounds=3
 /workflow:multi-cli-plan "Refactor payment module" --tools=gemini,codex,claude
 /workflow:multi-cli-plan "Fix memory leak" --mode=serial
-
-# Resume session
-/workflow:lite-execute --session=MCP-xxx
 ```
 
 **Context Source**: ACE semantic search + Multi-CLI analysis
 **Output Directory**: `.workflow/.multi-cli-plan/{session-id}/`
 **Default Max Rounds**: 3 (convergence may complete earlier)
 **CLI Tools**: @cli-discuss-agent (analysis), @cli-lite-planning-agent (plan generation)
+**Execution**: Auto-hands off to `/workflow:lite-execute --in-memory` after plan approval
 
 ## What & Why
 
@@ -71,22 +69,27 @@ Phase 3: Present Options
    └─ Display solutions with trade-offs from agent output
 
 Phase 4: User Decision
-   ├─ Approve solution → Phase 5
-   ├─ Need clarification → Return to Phase 2
-   └─ Change direction → Reset with feedback
+   ├─ Select solution approach
+   ├─ Select execution method (Agent/Codex/Auto)
+   ├─ Select code review tool (Skip/Gemini/Codex/Agent)
+   └─ Route:
+      ├─ Approve → Phase 5
+      ├─ Need More Analysis → Return to Phase 2
+      └─ Cancel → Save session
 
-Phase 5: Plan Generation (via @cli-lite-planning-agent)
-   ├─ Generate IMPL_PLAN.md + plan.json
-   └─ Hand off to /workflow:lite-execute
+Phase 5: Plan Generation & Execution Handoff
+   ├─ Generate plan.json (via @cli-lite-planning-agent)
+   ├─ Build executionContext with user selections
+   └─ Execute to /workflow:lite-execute --in-memory
 ```
 
 ### Agent Roles
 
 | Agent | Responsibility |
 |-------|---------------|
-| **Orchestrator** | Session management, ACE context, user decisions, phase transitions |
+| **Orchestrator** | Session management, ACE context, user decisions, phase transitions, executionContext assembly |
 | **@cli-discuss-agent** | Multi-CLI execution (Gemini/Codex/Claude), cross-verification, solution synthesis, synthesis.json output |
-| **@cli-lite-planning-agent** | Task decomposition, IMPL_PLAN.md + plan.json generation |
+| **@cli-lite-planning-agent** | Task decomposition, plan.json generation following schema |
 
 ## Core Responsibilities
 
@@ -205,25 +208,49 @@ Disagreements: ${synthesis.cross_verification.disagreements.length}
 **Decision Options**:
 ```javascript
 AskUserQuestion({
-  questions: [{
-    question: "Which solution approach?",
-    header: "Solution",
-    options: solutions.map((s, i) => ({
-      label: `Option ${i+1}: ${s.name}`,
-      description: `${s.effort} effort, ${s.risk} risk`
-    })).concat([
-      { label: "Need More Analysis", description: "Return to Phase 2" }
-    ])
-  }]
+  questions: [
+    {
+      question: "Which solution approach?",
+      header: "Solution",
+      multiSelect: false,
+      options: solutions.map((s, i) => ({
+        label: `Option ${i+1}: ${s.name}`,
+        description: `${s.effort} effort, ${s.risk} risk`
+      })).concat([
+        { label: "Need More Analysis", description: "Return to Phase 2" }
+      ])
+    },
+    {
+      question: "Execution method:",
+      header: "Execution",
+      multiSelect: false,
+      options: [
+        { label: "Agent", description: "@code-developer agent" },
+        { label: "Codex", description: "codex CLI tool" },
+        { label: "Auto", description: "Auto-select based on complexity" }
+      ]
+    },
+    {
+      question: "Code review after execution?",
+      header: "Review",
+      multiSelect: false,
+      options: [
+        { label: "Skip", description: "No review" },
+        { label: "Gemini Review", description: "Gemini CLI tool" },
+        { label: "Codex Review", description: "codex review --uncommitted" },
+        { label: "Agent Review", description: "Current agent review" }
+      ]
+    }
+  ]
 })
 ```
 
 **Routing**:
-- Approve → Phase 5
+- Approve + execution method → Phase 5
 - Need More Analysis → Phase 2 with feedback
-- Add constraints → Collect details, then Phase 5
+- Cancel → Save session for resumption
 
-### Phase 5: Plan Generation
+### Phase 5: Plan Generation & Execution Handoff
 
 **Step 1: Build Context-Package** (Orchestrator responsibility):
 ```javascript
@@ -340,31 +367,63 @@ ${JSON.stringify(contextPackage, null, 2)}
 4. Use implementation_plan.tasks[] as task foundation
 5. Preserve task dependencies (depends_on) and execution_flow
 6. Expand tasks with detailed acceptance criteria
-7. Generate IMPL_PLAN.md documenting milestones and key_points
-8. Generate plan.json following schema exactly
+7. Generate plan.json following schema exactly
 
 ## Output
-- ${sessionFolder}/IMPL_PLAN.md
 - ${sessionFolder}/plan.json
 
 ## Completion Checklist
-- [ ] IMPL_PLAN.md documents approach, milestones, technical_concerns
 - [ ] plan.json preserves task dependencies from implementation_plan
 - [ ] Task execution order follows execution_flow
 - [ ] Key_points reflected in task descriptions
 - [ ] User constraints applied to implementation
 - [ ] Acceptance criteria are testable
+- [ ] Schema fields match plan-json-schema.json exactly
 `
 })
 ```
 
-**Hand off to Execution**:
+**Step 3: Build executionContext**:
 ```javascript
-if (userConfirms) {
-  SlashCommand("/workflow:lite-execute --in-memory")
+// After plan.json is generated by cli-lite-planning-agent
+const plan = JSON.parse(Read(`${sessionFolder}/plan.json`))
+
+// Build executionContext (same structure as lite-plan)
+executionContext = {
+  planObject: plan,
+  explorationsContext: null,  // Multi-CLI doesn't use exploration files
+  explorationAngles: [],      // No exploration angles
+  explorationManifest: null,  // No manifest
+  clarificationContext: null,  // Store user feedback from Phase 2 if exists
+  executionMethod: userSelection.execution_method,  // From Phase 4
+  codeReviewTool: userSelection.code_review_tool,   // From Phase 4
+  originalUserInput: taskDescription,
+
+  // Optional: Task-level executor assignments
+  executorAssignments: null,  // Could be enhanced in future
+
+  session: {
+    id: sessionId,
+    folder: sessionFolder,
+    artifacts: {
+      explorations: [],  // No explorations in multi-CLI workflow
+      explorations_manifest: null,
+      plan: `${sessionFolder}/plan.json`,
+      synthesis_rounds: Array.from({length: currentRound}, (_, i) =>
+        `${sessionFolder}/rounds/${i+1}/synthesis.json`
+      ),
+      context_package: `${sessionFolder}/context-package.json`
+    }
+  }
 }
 ```
 
+**Step 4: Hand off to Execution**:
+```javascript
+// Execute to lite-execute with in-memory context
+SlashCommand("/workflow:lite-execute --in-memory")
+```
+
 ## Output File Structure
 
 ```
@@ -375,7 +434,6 @@ if (userConfirms) {
 │   ├── 2/synthesis.json        # Round 2 analysis (cli-discuss-agent)
 │   └── .../
 ├── context-package.json        # Extracted context for planning (orchestrator)
-├── IMPL_PLAN.md                # Documentation (cli-lite-planning-agent)
 └── plan.json                   # Structured plan (cli-lite-planning-agent)
 ```
 
@@ -386,8 +444,7 @@ if (userConfirms) {
 | `session-state.json` | Orchestrator | Session metadata, rounds, decisions |
 | `rounds/*/synthesis.json` | cli-discuss-agent | Solutions, convergence, cross-verification |
 | `context-package.json` | Orchestrator | Extracted solution, dependencies, consensus for planning |
-| `IMPL_PLAN.md` | cli-lite-planning-agent | Human-readable plan |
-| `plan.json` | cli-lite-planning-agent | Structured tasks for execution |
+| `plan.json` | cli-lite-planning-agent | Structured tasks for lite-execute |
 
 ## synthesis.json Schema
 
@@ -495,9 +552,6 @@ TodoWrite({ todos: [
 ## Related Commands
 
 ```bash
-# Resume saved session
-/workflow:lite-execute --session=MCP-xxx
-
 # Simpler single-round planning
 /workflow:lite-plan "task description"
 
@@ -505,6 +559,10 @@ TodoWrite({ todos: [
 /issue:discover-by-prompt "find issues"
 
 # View session files
-cat .workflow/.multi-cli-plan/{session-id}/IMPL_PLAN.md
+cat .workflow/.multi-cli-plan/{session-id}/plan.json
 cat .workflow/.multi-cli-plan/{session-id}/rounds/1/synthesis.json
+cat .workflow/.multi-cli-plan/{session-id}/context-package.json
+
+# Direct execution (if you have plan.json)
+/workflow:lite-execute plan.json
 ```

.claude/commands/workflow/tdd-plan.md

@@ -37,6 +37,44 @@ allowed-tools: SlashCommand(*), TodoWrite(*), Read(*), Bash(*)
 7. **Task Attachment Model**: SlashCommand execute **attaches** sub-tasks to current workflow. Orchestrator **executes** these attached tasks itself, then **collapses** them after completion
 8. **⚠️ CRITICAL: DO NOT STOP**: Continuous multi-phase workflow. After executing all attached tasks, immediately collapse them and execute next phase
 
+## TDD Compliance Requirements
+
+### The Iron Law
+
+```
+NO PRODUCTION CODE WITHOUT A FAILING TEST FIRST
+```
+
+**Enforcement Method**:
+- Phase 5: `implementation_approach` includes test-first steps (Red → Green → Refactor)
+- Green phase: Includes test-fix-cycle configuration (max 3 iterations)
+- Auto-revert: Triggered when max iterations reached without passing tests
+
+**Verification**: Phase 6 validates Red-Green-Refactor structure in all generated tasks
+
+### TDD Compliance Checkpoint
+
+| Checkpoint | Validation Phase | Evidence Required |
+|------------|------------------|-------------------|
+| Test-first structure | Phase 5 | `implementation_approach` has 3 steps |
+| Red phase exists | Phase 6 | Step 1: `tdd_phase: "red"` |
+| Green phase with test-fix | Phase 6 | Step 2: `tdd_phase: "green"` + test-fix-cycle |
+| Refactor phase exists | Phase 6 | Step 3: `tdd_phase: "refactor"` |
+
+### Core TDD Principles (from ref skills)
+
+**Red Flags - STOP and Reassess**:
+- Code written before test
+- Test passes immediately (no Red phase witnessed)
+- Cannot explain why test should fail
+- "Just this once" rationalization
+- "Tests after achieve same goals" thinking
+
+**Why Order Matters**:
+- Tests written after code pass immediately → proves nothing
+- Test-first forces edge case discovery before implementation
+- Tests-after verify what was built, not what's required
+
 ## 6-Phase Execution (with Conflict Resolution)
 
 ### Phase 1: Session Discovery
@@ -183,7 +221,7 @@ SlashCommand(command="/workflow:tools:conflict-resolution --session [sessionId]
   {"content": "Phase 3: Test Coverage Analysis", "status": "completed", "activeForm": "Executing test coverage analysis"},
   {"content": "Phase 4: Conflict Resolution", "status": "in_progress", "activeForm": "Executing conflict resolution"},
   {"content": "  → Detect conflicts with CLI analysis", "status": "in_progress", "activeForm": "Detecting conflicts"},
-  {"content": "  → Present conflicts to user", "status": "pending", "activeForm": "Presenting conflicts"},
+  {"content": "  → Log and analyze detected conflicts", "status": "pending", "activeForm": "Analyzing conflicts"},
   {"content": "  → Apply resolution strategies", "status": "pending", "activeForm": "Applying resolution strategies"},
   {"content": "Phase 5: TDD Task Generation", "status": "pending", "activeForm": "Executing TDD task generation"},
   {"content": "Phase 6: TDD Structure Validation", "status": "pending", "activeForm": "Validating TDD structure"}
@@ -251,6 +289,13 @@ SlashCommand(command="/workflow:tools:task-generate-tdd --session [sessionId]")
 - IMPL_PLAN.md contains workflow_type: "tdd" in frontmatter
 - Task count ≤10 (compliance with task limit)
 
+**Red Flag Detection** (Non-Blocking Warnings):
+- Task count >10: `⚠️ High task count may indicate insufficient decomposition`
+- Missing test-fix-cycle: `⚠️ Green phase lacks auto-revert configuration`
+- Generic task names: `⚠️ Vague task names suggest unclear TDD cycles`
+
+**Action**: Log warnings to `.workflow/active/[sessionId]/.process/tdd-warnings.log` (non-blocking)
+
 <!-- TodoWrite: When task-generate-tdd executed, INSERT 3 task-generate-tdd tasks -->
 
 **TodoWrite Update (Phase 5 SlashCommand executed - tasks attached)**:
@@ -302,6 +347,42 @@ SlashCommand(command="/workflow:tools:task-generate-tdd --session [sessionId]")
 5. Test-fix cycle: Green phase step includes test-fix-cycle logic with max_iterations
 6. Task count: Total tasks ≤10 (simple + subtasks)
 
+**Red Flag Checklist** (from TDD best practices):
+- [ ] No tasks skip Red phase (`tdd_phase: "red"` exists in step 1)
+- [ ] Test files referenced in Red phase (explicit paths, not placeholders)
+- [ ] Green phase has test-fix-cycle with `max_iterations` configured
+- [ ] Refactor phase has clear completion criteria
+
+**Non-Compliance Warning Format**:
+```
+⚠️ TDD Red Flag: [issue description]
+   Task: [IMPL-N]
+   Recommendation: [action to fix]
+```
+
+**Evidence Gathering** (Before Completion Claims):
+
+```bash
+# Verify session artifacts exist
+ls -la .workflow/active/[sessionId]/{IMPL_PLAN.md,TODO_LIST.md}
+ls -la .workflow/active/[sessionId]/.task/IMPL-*.json
+
+# Count generated artifacts
+echo "IMPL tasks: $(ls .workflow/active/[sessionId]/.task/IMPL-*.json 2>/dev/null | wc -l)"
+
+# Sample task structure verification (first task)
+jq '{id, tdd: .meta.tdd_workflow, phases: [.flow_control.implementation_approach[].tdd_phase]}' \
+  "$(ls .workflow/active/[sessionId]/.task/IMPL-*.json | head -1)"
+```
+
+**Evidence Required Before Summary**:
+| Evidence Type | Verification Method | Pass Criteria |
+|---------------|---------------------|---------------|
+| File existence | `ls -la` artifacts | All files present |
+| Task count | Count IMPL-*.json | Count matches claims |
+| TDD structure | jq sample extraction | Shows red/green/refactor |
+| Warning log | Check tdd-warnings.log | Logged (may be empty) |
+
 **Return Summary**:
 ```
 TDD Planning complete for session: [sessionId]
@@ -333,6 +414,9 @@ TDD Configuration:
 - Green phase includes test-fix cycle (max 3 iterations)
 - Auto-revert on max iterations reached
 
+⚠️ ACTION REQUIRED: Before execution, ensure you understand WHY each Red phase test is expected to fail.
+   This is crucial for valid TDD - if you don't know why the test fails, you can't verify it tests the right thing.
+
 Recommended Next Steps:
 1. /workflow:action-plan-verify --session [sessionId]  # Verify TDD plan quality and dependencies
 2. /workflow:execute --session [sessionId]  # Start TDD execution
@@ -400,7 +484,7 @@ TDD Workflow Orchestrator
 │  IF conflict_risk ≥ medium:
 │  └─ /workflow:tools:conflict-resolution             ← ATTACHED (3 tasks)
 │     ├─ Phase 4.1: Detect conflicts with CLI
-│     ├─ Phase 4.2: Present conflicts to user
+│     ├─ Phase 4.2: Log and analyze detected conflicts
 │     └─ Phase 4.3: Apply resolution strategies
 │     └─ Returns: conflict-resolution.json            ← COLLAPSED
 │  ELSE:
@@ -439,6 +523,34 @@ Convert user input to TDD-structured format:
 - **Command failure**: Keep phase in_progress, report error
 - **TDD validation failure**: Report incomplete chains or wrong dependencies
 
+### TDD Warning Patterns
+
+| Pattern | Warning Message | Recommended Action |
+|---------|----------------|-------------------|
+| Task count >10 | High task count detected | Consider splitting into multiple sessions |
+| Missing test-fix-cycle | Green phase lacks auto-revert | Add `max_iterations: 3` to task config |
+| Red phase missing test path | Test file path not specified | Add explicit test file paths |
+| Generic task names | Vague names like "Add feature" | Use specific behavior descriptions |
+| No refactor criteria | Refactor phase lacks completion criteria | Define clear refactor scope |
+
+### Non-Blocking Warning Policy
+
+**All warnings are advisory** - they do not halt execution:
+1. Warnings logged to `.process/tdd-warnings.log`
+2. Summary displayed in Phase 6 output
+3. User decides whether to address before `/workflow:execute`
+
+### Error Handling Quick Reference
+
+| Error Type | Detection | Recovery Action |
+|------------|-----------|-----------------|
+| Parsing failure | Empty/malformed output | Retry once, then report |
+| Missing context-package | File read error | Re-run `/workflow:tools:context-gather` |
+| Invalid task JSON | jq parse error | Report malformed file path |
+| High task count (>10) | Count validation | Log warning, continue (non-blocking) |
+| Test-context missing | File not found | Re-run `/workflow:tools:test-context-gather` |
+| Phase timeout | No response | Retry phase, check CLI connectivity |
+
 ## Related Commands
 
 **Prerequisite Commands**:
@@ -458,3 +570,28 @@ Convert user input to TDD-structured format:
 - `/workflow:execute` - Begin TDD implementation
 - `/workflow:tdd-verify` - Post-execution: Verify TDD compliance and generate quality report
 
+## Next Steps Decision Table
+
+| Situation | Recommended Command | Purpose |
+|-----------|---------------------|---------|
+| First time planning | `/workflow:action-plan-verify` | Validate task structure before execution |
+| Warnings in tdd-warnings.log | Review log, refine tasks | Address Red Flags before proceeding |
+| High task count warning | Consider `/workflow:session:start` | Split into focused sub-sessions |
+| Ready to implement | `/workflow:execute` | Begin TDD Red-Green-Refactor cycles |
+| After implementation | `/workflow:tdd-verify` | Generate TDD compliance report |
+| Need to review tasks | `/workflow:status --session [id]` | Inspect current task breakdown |
+| Plan needs changes | `/task:replan` | Update task JSON with new requirements |
+
+### TDD Workflow State Transitions
+
+```
+/workflow:tdd-plan
+        ↓
+[Planning Complete] ──→ /workflow:action-plan-verify (recommended)
+        ↓
+[Verified/Ready] ─────→ /workflow:execute
+        ↓
+[Implementation] ─────→ /workflow:tdd-verify (post-execution)
+        ↓
+[Quality Report] ─────→ Done or iterate
+```

.claude/commands/workflow/tools/conflict-resolution.md

@@ -154,8 +154,8 @@ Task(subagent_type="cli-execution-agent", run_in_background=false, prompt=`
     - Validation of exploration conflict_indicators
     - ModuleOverlap conflicts with overlap_analysis
     - Targeted clarification questions
-  RULES: $(cat ~/.claude/workflows/cli-templates/prompts/analysis/02-analyze-code-patterns.txt) | Focus on breaking changes, migration needs, and functional overlaps | Prioritize exploration-identified conflicts | analysis=READ-ONLY
-  " --tool gemini --mode analysis --cd {project_root}
+  CONSTRAINTS: Focus on breaking changes, migration needs, and functional overlaps | Prioritize exploration-identified conflicts | analysis=READ-ONLY
+  " --tool gemini --mode analysis --rule analysis-code-patterns --cd {project_root}
 
   Fallback: Qwen (same prompt) → Claude (manual analysis)
 

.claude/commands/workflow/tools/test-concept-enhanced.md

@@ -90,7 +90,7 @@ Template: ~/.claude/workflows/cli-templates/prompts/test/test-concept-analysis.t
 
 ## EXECUTION STEPS
 1. Execute Gemini analysis:
-   ccw cli -p "$(cat ~/.claude/workflows/cli-templates/prompts/test/test-concept-analysis.txt)" --tool gemini --mode write --cd .workflow/active/{test_session_id}/.process
+   ccw cli -p "..." --tool gemini --mode write --rule test-test-concept-analysis --cd .workflow/active/{test_session_id}/.process
 
 2. Generate TEST_ANALYSIS_RESULTS.md:
    Synthesize gemini-test-analysis.md into standardized format for task generation

.claude/skills/ccw-loop/README.md

@@ -0,0 +1,303 @@
+# CCW Loop Skill
+
+无状态迭代开发循环工作流，支持开发 (Develop)、调试 (Debug)、验证 (Validate) 三个阶段，每个阶段都有独立的文件记录进展。
+
+## Overview
+
+CCW Loop 是一个自主模式 (Autonomous) 的 Skill，通过文件驱动的无状态循环，帮助开发者系统化地完成开发任务。
+
+### 核心特性
+
+1. **无状态循环**: 每次执行从文件读取状态，不依赖内存
+2. **文件驱动**: 所有进度记录在 Markdown 文件中，可审计、可回顾
+3. **Gemini 辅助**: 关键决策点使用 CLI 工具进行深度分析
+4. **可恢复**: 任何时候中断后可继续
+5. **双模式**: 支持交互式和自动循环
+
+### 三大阶段
+
+- **Develop**: 任务分解 → 代码实现 → 进度记录
+- **Debug**: 假设生成 → 证据收集 → 根因分析 → 修复验证
+- **Validate**: 测试执行 → 覆盖率检查 → 质量评估
+
+## Installation
+
+已包含在 `.claude/skills/ccw-loop/`，无需额外安装。
+
+## Usage
+
+### 基本用法
+
+```bash
+# 启动新循环
+/ccw-loop "实现用户认证功能"
+
+# 继续现有循环
+/ccw-loop --resume LOOP-auth-2026-01-22
+
+# 自动循环模式
+/ccw-loop --auto "修复登录bug并添加测试"
+```
+
+### 交互式流程
+
+```
+1. 启动: /ccw-loop "任务描述"
+2. 初始化: 自动分析任务并生成子任务列表
+3. 显示菜单:
+   - 📝 继续开发 (Develop)
+   - 🔍 开始调试 (Debug)
+   - ✅ 运行验证 (Validate)
+   - 📊 查看详情 (Status)
+   - 🏁 完成循环 (Complete)
+   - 🚪 退出 (Exit)
+4. 执行选择的动作
+5. 重复步骤 3-4 直到完成
+```
+
+### 自动循环流程
+
+```
+Develop (所有任务) → Debug (如有需要) → Validate → 完成
+```
+
+## Directory Structure
+
+```
+.workflow/.loop/{session-id}/
+├── meta.json              # 会话元数据 (不可修改)
+├── state.json             # 当前状态 (每次更新)
+├── summary.md             # 完成报告 (结束时生成)
+├── develop/
+│   ├── progress.md        # 开发进度时间线
+│   ├── tasks.json         # 任务列表
+│   └── changes.log        # 代码变更日志 (NDJSON)
+├── debug/
+│   ├── understanding.md   # 理解演变文档
+│   ├── hypotheses.json    # 假设历史
+│   └── debug.log          # 调试日志 (NDJSON)
+└── validate/
+    ├── validation.md      # 验证报告
+    ├── test-results.json  # 测试结果
+    └── coverage.json      # 覆盖率数据
+```
+
+## Action Reference
+
+| Action | 描述 | 触发条件 |
+|--------|------|----------|
+| action-init | 初始化会话 | 首次启动 |
+| action-menu | 显示操作菜单 | 交互模式下每次循环 |
+| action-develop-with-file | 执行开发任务 | 有待处理任务 |
+| action-debug-with-file | 假设驱动调试 | 需要调试 |
+| action-validate-with-file | 运行测试验证 | 需要验证 |
+| action-complete | 完成并生成报告 | 所有任务完成 |
+
+详细说明见 [specs/action-catalog.md](specs/action-catalog.md)
+
+## CLI Integration
+
+CCW Loop 在关键决策点集成 CLI 工具:
+
+### 任务分解 (action-init)
+```bash
+ccw cli -p "PURPOSE: 分解开发任务..."
+  --tool gemini
+  --mode analysis
+  --rule planning-breakdown-task-steps
+```
+
+### 代码实现 (action-develop)
+```bash
+ccw cli -p "PURPOSE: 实现功能代码..."
+  --tool gemini
+  --mode write
+  --rule development-implement-feature
+```
+
+### 假设生成 (action-debug - 探索)
+```bash
+ccw cli -p "PURPOSE: Generate debugging hypotheses..."
+  --tool gemini
+  --mode analysis
+  --rule analysis-diagnose-bug-root-cause
+```
+
+### 证据分析 (action-debug - 分析)
+```bash
+ccw cli -p "PURPOSE: Analyze debug log evidence..."
+  --tool gemini
+  --mode analysis
+  --rule analysis-diagnose-bug-root-cause
+```
+
+### 质量评估 (action-validate)
+```bash
+ccw cli -p "PURPOSE: Analyze test results and coverage..."
+  --tool gemini
+  --mode analysis
+  --rule analysis-review-code-quality
+```
+
+## State Management
+
+### State Schema
+
+参见 [phases/state-schema.md](phases/state-schema.md)
+
+### State Transitions
+
+```
+pending → running → completed
+              ↓
+         user_exit
+              ↓
+            failed
+```
+
+### State Recovery
+
+如果 `state.json` 损坏，可从其他文件重建:
+- develop/tasks.json → develop.*
+- debug/hypotheses.json → debug.*
+- validate/test-results.json → validate.*
+
+## Examples
+
+### Example 1: 功能开发
+
+```bash
+# 1. 启动循环
+/ccw-loop "Add user profile page"
+
+# 2. 系统初始化，生成任务:
+#    - task-001: Create profile component
+#    - task-002: Add API endpoints
+#    - task-003: Implement tests
+
+# 3. 选择 "继续开发"
+#    → 执行 task-001 (Gemini 辅助实现)
+#    → 更新 progress.md
+
+# 4. 重复开发直到所有任务完成
+
+# 5. 选择 "运行验证"
+#    → 运行测试
+#    → 检查覆盖率
+#    → 生成 validation.md
+
+# 6. 选择 "完成循环"
+#    → 生成 summary.md
+#    → 询问是否扩展为 Issue
+```
+
+### Example 2: Bug 修复
+
+```bash
+# 1. 启动循环
+/ccw-loop "Fix login timeout issue"
+
+# 2. 选择 "开始调试"
+#    → 输入 bug 描述: "Login times out after 30s"
+#    → Gemini 生成假设 (H1, H2, H3)
+#    → 添加 NDJSON 日志
+#    → 提示复现 bug
+
+# 3. 复现 bug (在应用中操作)
+
+# 4. 再次选择 "开始调试"
+#    → 解析 debug.log
+#    → Gemini 分析证据
+#    → H2 确认为根因
+#    → 生成修复代码
+#    → 更新 understanding.md
+
+# 5. 选择 "运行验证"
+#    → 测试通过
+
+# 6. 完成
+```
+
+## Templates
+
+- [progress-template.md](templates/progress-template.md): 开发进度文档模板
+- [understanding-template.md](templates/understanding-template.md): 调试理解文档模板
+- [validation-template.md](templates/validation-template.md): 验证报告模板
+
+## Specifications
+
+- [loop-requirements.md](specs/loop-requirements.md): 循环需求规范
+- [action-catalog.md](specs/action-catalog.md): 动作目录
+
+## Integration
+
+### Dashboard Integration
+
+CCW Loop 与 Dashboard Loop Monitor 集成:
+- Dashboard 创建 Loop → 触发此 Skill
+- state.json → Dashboard 实时显示
+- 任务列表双向同步
+- 控制按钮映射到 actions
+
+### Issue System Integration
+
+完成后可扩展为 Issue:
+- 维度: test, enhance, refactor, doc
+- 自动调用 `/issue:new`
+- 上下文自动填充
+
+## Error Handling
+
+| 情况 | 处理 |
+|------|------|
+| Session 不存在 | 创建新会话 |
+| state.json 损坏 | 从文件重建 |
+| CLI 工具失败 | 回退到手动模式 |
+| 测试失败 | 循环回到 develop/debug |
+| >10 迭代 | 警告用户，建议拆分 |
+
+## Limitations
+
+1. **单会话限制**: 同一时间只能有一个活跃会话
+2. **迭代限制**: 建议不超过 10 次迭代
+3. **CLI 依赖**: 部分功能依赖 Gemini CLI 可用性
+4. **测试框架**: 需要 package.json 中定义测试脚本
+
+## Troubleshooting
+
+### Q: 如何查看当前会话状态？
+
+A: 在菜单中选择 "查看详情 (Status)"
+
+### Q: 如何恢复中断的会话？
+
+A: 使用 `--resume` 参数:
+```bash
+/ccw-loop --resume LOOP-xxx-2026-01-22
+```
+
+### Q: 如果 CLI 工具失败怎么办？
+
+A: Skill 会自动降级到手动模式，提示用户手动输入
+
+### Q: 如何添加自定义 action？
+
+A: 参见 [specs/action-catalog.md](specs/action-catalog.md) 的 "Action Extensions" 部分
+
+## Contributing
+
+添加新功能:
+1. 创建 action 文件在 `phases/actions/`
+2. 更新 orchestrator 决策逻辑
+3. 添加到 action-catalog.md
+4. 更新 action-menu.md
+
+## License
+
+MIT
+
+---
+
+**Version**: 1.0.0
+**Last Updated**: 2026-01-22
+**Author**: CCW Team

.claude/skills/ccw-loop/SKILL.md

@@ -0,0 +1,259 @@
+---
+name: ccw-loop
+description: Stateless iterative development loop workflow with documented progress. Supports develop, debug, and validate phases with file-based state tracking. Triggers on "ccw-loop", "dev loop", "development loop", "开发循环", "迭代开发".
+allowed-tools: Task(*), AskUserQuestion(*), Read(*), Grep(*), Glob(*), Bash(*), Edit(*), Write(*), TodoWrite(*)
+---
+
+# CCW Loop - Stateless Iterative Development Workflow
+
+无状态迭代开发循环工作流，支持开发 (develop)、调试 (debug)、验证 (validate) 三个阶段，每个阶段都有独立的文件记录进展。
+
+## Arguments
+
+| Arg | Required | Description |
+|-----|----------|-------------|
+| task | No | Task description (for new loop, mutually exclusive with --loop-id) |
+| --loop-id | No | Existing loop ID to continue (from API or previous session) |
+| --auto | No | Auto-cycle mode (develop → debug → validate → complete) |
+
+## Unified Architecture (API + Skill Integration)
+
+```
+┌─────────────────────────────────────────────────────────────────┐
+│                     Dashboard (UI)                              │
+│  [Create] [Start] [Pause] [Resume] [Stop] [View Progress]       │
+└─────────────────────────────────────────────────────────────────┘
+                              │
+                              ▼
+┌─────────────────────────────────────────────────────────────────┐
+│              loop-v2-routes.ts (Control Plane)                  │
+│                                                                 │
+│  State: .loop/{loopId}.json (MASTER)                            │
+│  Tasks: .loop/{loopId}.tasks.jsonl                              │
+│                                                                 │
+│  /start → Trigger ccw-loop skill with --loop-id                 │
+│  /pause → Set status='paused' (skill checks before action)      │
+│  /stop  → Set status='failed' (skill terminates)                │
+│  /resume → Set status='running' (skill continues)               │
+└─────────────────────────────────────────────────────────────────┘
+                              │
+                              ▼
+┌─────────────────────────────────────────────────────────────────┐
+│               ccw-loop Skill (Execution Plane)                  │
+│                                                                 │
+│  Reads/Writes: .loop/{loopId}.json (unified state)              │
+│  Writes: .loop/{loopId}.progress/* (progress files)             │
+│                                                                 │
+│  BEFORE each action:                                            │
+│    → Check status: paused/stopped → exit gracefully             │
+│    → running → continue with action                             │
+│                                                                 │
+│  Actions: init → develop → debug → validate → complete          │
+└─────────────────────────────────────────────────────────────────┘
+```
+
+## Key Design Principles
+
+1. **统一状态**: API 和 Skill 共享 `.loop/{loopId}.json` 状态文件
+2. **控制信号**: Skill 每个 Action 前检查 status 字段 (paused/stopped)
+3. **文件驱动**: 所有进度、理解、结果都记录在 `.loop/{loopId}.progress/`
+4. **可恢复**: 任何时候可以继续之前的循环 (`--loop-id`)
+5. **双触发**: 支持 API 触发 (`--loop-id`) 和直接调用 (task description)
+6. **Gemini 辅助**: 使用 CLI 工具进行深度分析和假设验证
+
+## Execution Modes
+
+### Mode 1: Interactive (交互式)
+
+用户手动选择每个动作，适合复杂任务。
+
+```
+用户 → 选择动作 → 执行 → 查看结果 → 选择下一动作
+```
+
+### Mode 2: Auto-Loop (自动循环)
+
+按预设顺序自动执行，适合标准开发流程。
+
+```
+Develop → Debug → Validate → (如有问题) → Develop → ...
+```
+
+## Session Structure (Unified Location)
+
+```
+.loop/
+├── {loopId}.json              # 主状态文件 (API + Skill 共享)
+├── {loopId}.tasks.jsonl       # 任务列表 (API 管理)
+└── {loopId}.progress/         # Skill 进度文件
+    ├── develop.md             # 开发进度记录
+    ├── debug.md               # 理解演变文档
+    ├── validate.md            # 验证报告
+    ├── changes.log            # 代码变更日志 (NDJSON)
+    └── debug.log              # 调试日志 (NDJSON)
+```
+
+## Directory Setup
+
+```javascript
+// loopId 来源:
+// 1. API 触发时: 从 --loop-id 参数获取
+// 2. 直接调用时: 生成新的 loop-v2-{timestamp}-{random}
+
+const loopId = args['--loop-id'] || generateLoopId()
+const loopFile = `.loop/${loopId}.json`
+const progressDir = `.loop/${loopId}.progress`
+
+// 创建进度目录
+Bash(`mkdir -p "${progressDir}"`)
+```
+
+## Action Catalog
+
+| Action | Purpose | Output Files | CLI Integration |
+|--------|---------|--------------|-----------------|
+| [action-init](phases/actions/action-init.md) | 初始化循环会话 | meta.json, state.json | - |
+| [action-develop-with-file](phases/actions/action-develop-with-file.md) | 开发任务执行 | progress.md, tasks.json | gemini --mode write |
+| [action-debug-with-file](phases/actions/action-debug-with-file.md) | 假设驱动调试 | understanding.md, hypotheses.json | gemini --mode analysis |
+| [action-validate-with-file](phases/actions/action-validate-with-file.md) | 测试与验证 | validation.md, test-results.json | gemini --mode analysis |
+| [action-complete](phases/actions/action-complete.md) | 完成循环 | summary.md | - |
+| [action-menu](phases/actions/action-menu.md) | 显示操作菜单 | - | - |
+
+## Usage
+
+```bash
+# 启动新循环 (直接调用)
+/ccw-loop "实现用户认证功能"
+
+# 继续现有循环 (API 触发或手动恢复)
+/ccw-loop --loop-id loop-v2-20260122-abc123
+
+# 自动循环模式
+/ccw-loop --auto "修复登录bug并添加测试"
+
+# API 触发自动循环
+/ccw-loop --loop-id loop-v2-20260122-abc123 --auto
+```
+
+## Execution Flow
+
+```
+┌─────────────────────────────────────────────────────────────────┐
+│  /ccw-loop [<task> | --loop-id <id>] [--auto]                    │
+├─────────────────────────────────────────────────────────────────┤
+│                                                                  │
+│  1. Parameter Detection:                                         │
+│     ├─ IF --loop-id provided:                                    │
+│     │   ├─ Read .loop/{loopId}.json                              │
+│     │   ├─ Validate status === 'running'                         │
+│     │   └─ Continue from skill_state.current_action              │
+│     └─ ELSE (task description):                                  │
+│         ├─ Generate new loopId                                   │
+│         ├─ Create .loop/{loopId}.json                            │
+│         └─ Initialize with action-init                           │
+│                                                                  │
+│  2. Orchestrator Loop:                                           │
+│     ├─ Read state from .loop/{loopId}.json                       │
+│     ├─ Check control signals:                                    │
+│     │   ├─ status === 'paused' → Exit (wait for resume)          │
+│     │   ├─ status === 'failed' → Exit with error                 │
+│     │   └─ status === 'running' → Continue                       │
+│     ├─ Show menu / auto-select next action                       │
+│     ├─ Execute action                                            │
+│     ├─ Update .loop/{loopId}.progress/{action}.md                │
+│     ├─ Update .loop/{loopId}.json (skill_state)                  │
+│     └─ Loop or exit based on user choice / completion            │
+│                                                                  │
+│  3. Action Execution:                                            │
+│     ├─ BEFORE: checkControlSignals() → exit if paused/stopped    │
+│     ├─ Develop: Plan → Implement → Document progress             │
+│     ├─ Debug: Hypothesize → Instrument → Analyze → Fix           │
+│     ├─ Validate: Test → Check → Report                           │
+│     └─ AFTER: Update skill_state in .loop/{loopId}.json          │
+│                                                                  │
+│  4. Termination:                                                 │
+│     ├─ Control signal: paused (graceful exit, wait resume)       │
+│     ├─ Control signal: stopped (failed state)                    │
+│     ├─ User exits (interactive mode)                             │
+│     ├─ All tasks completed (status → completed)                  │
+│     └─ Max iterations reached                                    │
+│                                                                  │
+└─────────────────────────────────────────────────────────────────┘
+```
+
+## Reference Documents
+
+| Document | Purpose |
+|----------|---------|
+| [phases/orchestrator.md](phases/orchestrator.md) | 编排器：状态读取 + 动作选择 |
+| [phases/state-schema.md](phases/state-schema.md) | 状态结构定义 |
+| [specs/loop-requirements.md](specs/loop-requirements.md) | 循环需求规范 |
+| [specs/action-catalog.md](specs/action-catalog.md) | 动作目录 |
+| [templates/progress-template.md](templates/progress-template.md) | 进度文档模板 |
+| [templates/understanding-template.md](templates/understanding-template.md) | 理解文档模板 |
+
+## Integration with Loop Monitor (Dashboard)
+
+此 Skill 与 CCW Dashboard 的 Loop Monitor 实现 **控制平面 + 执行平面** 分离架构：
+
+### Control Plane (Dashboard/API → loop-v2-routes.ts)
+
+1. **创建循环**: `POST /api/loops/v2` → 创建 `.loop/{loopId}.json`
+2. **启动执行**: `POST /api/loops/v2/:loopId/start` → 触发 `/ccw-loop --loop-id {loopId} --auto`
+3. **暂停执行**: `POST /api/loops/v2/:loopId/pause` → 设置 `status='paused'` (Skill 下次检查时退出)
+4. **恢复执行**: `POST /api/loops/v2/:loopId/resume` → 设置 `status='running'` → 重新触发 Skill
+5. **停止执行**: `POST /api/loops/v2/:loopId/stop` → 设置 `status='failed'`
+
+### Execution Plane (ccw-loop Skill)
+
+1. **读取状态**: 从 `.loop/{loopId}.json` 读取 API 设置的状态
+2. **检查控制**: 每个 Action 前检查 `status` 字段
+3. **执行动作**: develop → debug → validate → complete
+4. **更新进度**: 写入 `.loop/{loopId}.progress/*.md` 和更新 `skill_state`
+5. **状态同步**: Dashboard 通过读取 `.loop/{loopId}.json` 获取进度
+
+## CLI Integration Points
+
+### Develop Phase
+```bash
+ccw cli -p "PURPOSE: Implement {task}...
+TASK: • Analyze requirements • Write code • Update progress
+MODE: write
+CONTEXT: @progress.md @tasks.json
+EXPECTED: Implementation + updated progress.md
+" --tool gemini --mode write --rule development-implement-feature
+```
+
+### Debug Phase
+```bash
+ccw cli -p "PURPOSE: Generate debugging hypotheses...
+TASK: • Analyze error • Generate hypotheses • Add instrumentation
+MODE: analysis
+CONTEXT: @understanding.md @debug.log
+EXPECTED: Hypotheses + instrumentation plan
+" --tool gemini --mode analysis --rule analysis-diagnose-bug-root-cause
+```
+
+### Validate Phase
+```bash
+ccw cli -p "PURPOSE: Validate implementation...
+TASK: • Run tests • Check coverage • Verify requirements
+MODE: analysis
+CONTEXT: @validation.md @test-results.json
+EXPECTED: Validation report
+" --tool gemini --mode analysis --rule analysis-review-code-quality
+```
+
+## Error Handling
+
+| Situation | Action |
+|-----------|--------|
+| Session not found | Create new session |
+| State file corrupted | Rebuild from file contents |
+| CLI tool fails | Fallback to manual analysis |
+| Tests fail | Loop back to develop/debug |
+| >10 iterations | Warn user, suggest break |
+
+## Post-Completion Expansion
+
+完成后询问用户是否扩展为 issue (test/enhance/refactor/doc)，选中项调用 `/issue:new "{summary} - {dimension}"`

.claude/skills/ccw-loop/phases/actions/action-complete.md

@@ -0,0 +1,320 @@
+# Action: Complete
+
+完成 CCW Loop 会话，生成总结报告。
+
+## Purpose
+
+- 生成完成报告
+- 汇总所有阶段成果
+- 提供后续建议
+- 询问是否扩展为 Issue
+
+## Preconditions
+
+- [ ] state.initialized === true
+- [ ] state.status === 'running'
+
+## Execution
+
+### Step 1: 汇总统计
+
+```javascript
+const getUtc8ISOString = () => new Date(Date.now() + 8 * 60 * 60 * 1000).toISOString()
+
+const sessionFolder = `.workflow/.loop/${state.session_id}`
+
+const stats = {
+  // 时间统计
+  duration: Date.now() - new Date(state.created_at).getTime(),
+  iterations: state.iteration_count,
+
+  // 开发统计
+  develop: {
+    total_tasks: state.develop.total_count,
+    completed_tasks: state.develop.completed_count,
+    completion_rate: state.develop.total_count > 0
+      ? (state.develop.completed_count / state.develop.total_count * 100).toFixed(1)
+      : 0
+  },
+
+  // 调试统计
+  debug: {
+    iterations: state.debug.iteration,
+    hypotheses_tested: state.debug.hypotheses.length,
+    root_cause_found: state.debug.confirmed_hypothesis !== null
+  },
+
+  // 验证统计
+  validate: {
+    runs: state.validate.test_results.length,
+    passed: state.validate.passed,
+    coverage: state.validate.coverage,
+    failed_tests: state.validate.failed_tests.length
+  }
+}
+
+console.log('\n生成完成报告...')
+```
+
+### Step 2: 生成总结报告
+
+```javascript
+const summaryReport = `# CCW Loop Session Summary
+
+**Session ID**: ${state.session_id}
+**Task**: ${state.task_description}
+**Started**: ${state.created_at}
+**Completed**: ${getUtc8ISOString()}
+**Duration**: ${formatDuration(stats.duration)}
+
+---
+
+## Executive Summary
+
+${state.validate.passed
+  ? '✅ **任务成功完成** - 所有测试通过，验证成功'
+  : state.develop.completed_count === state.develop.total_count
+    ? '⚠️ **开发完成，验证未通过** - 需要进一步调试'
+    : '⏸️ **任务部分完成** - 仍有待处理项'}
+
+---
+
+## Development Phase
+
+| Metric | Value |
+|--------|-------|
+| Total Tasks | ${stats.develop.total_tasks} |
+| Completed | ${stats.develop.completed_tasks} |
+| Completion Rate | ${stats.develop.completion_rate}% |
+
+### Completed Tasks
+
+${state.develop.tasks.filter(t => t.status === 'completed').map(t => `
+- ✅ ${t.description}
+  - Files: ${t.files_changed?.join(', ') || 'N/A'}
+  - Completed: ${t.completed_at}
+`).join('\n')}
+
+### Pending Tasks
+
+${state.develop.tasks.filter(t => t.status !== 'completed').map(t => `
+- ⏳ ${t.description}
+`).join('\n') || '_None_'}
+
+---
+
+## Debug Phase
+
+| Metric | Value |
+|--------|-------|
+| Iterations | ${stats.debug.iterations} |
+| Hypotheses Tested | ${stats.debug.hypotheses_tested} |
+| Root Cause Found | ${stats.debug.root_cause_found ? 'Yes' : 'No'} |
+
+${stats.debug.root_cause_found ? `
+### Confirmed Root Cause
+
+**${state.debug.confirmed_hypothesis}**: ${state.debug.hypotheses.find(h => h.id === state.debug.confirmed_hypothesis)?.description || 'N/A'}
+` : ''}
+
+### Hypothesis Summary
+
+${state.debug.hypotheses.map(h => `
+- **${h.id}**: ${h.status.toUpperCase()}
+  - ${h.description}
+`).join('\n') || '_No hypotheses tested_'}
+
+---
+
+## Validation Phase
+
+| Metric | Value |
+|--------|-------|
+| Test Runs | ${stats.validate.runs} |
+| Status | ${stats.validate.passed ? 'PASSED' : 'FAILED'} |
+| Coverage | ${stats.validate.coverage || 'N/A'}% |
+| Failed Tests | ${stats.validate.failed_tests} |
+
+${stats.validate.failed_tests > 0 ? `
+### Failed Tests
+
+${state.validate.failed_tests.map(t => `- ❌ ${t}`).join('\n')}
+` : ''}
+
+---
+
+## Files Modified
+
+${listModifiedFiles(sessionFolder)}
+
+---
+
+## Key Learnings
+
+${state.debug.iteration > 0 ? `
+### From Debugging
+
+${extractLearnings(state.debug.hypotheses)}
+` : ''}
+
+---
+
+## Recommendations
+
+${generateRecommendations(stats, state)}
+
+---
+
+## Session Artifacts
+
+| File | Description |
+|------|-------------|
+| \`develop/progress.md\` | Development progress timeline |
+| \`develop/tasks.json\` | Task list with status |
+| \`debug/understanding.md\` | Debug exploration and learnings |
+| \`debug/hypotheses.json\` | Hypothesis history |
+| \`validate/validation.md\` | Validation report |
+| \`validate/test-results.json\` | Test execution results |
+
+---
+
+*Generated by CCW Loop at ${getUtc8ISOString()}*
+`
+
+Write(`${sessionFolder}/summary.md`, summaryReport)
+console.log(`\n报告已保存: ${sessionFolder}/summary.md`)
+```
+
+### Step 3: 询问后续扩展
+
+```javascript
+console.log('\n' + '═'.repeat(60))
+console.log('  任务已完成')
+console.log('═'.repeat(60))
+
+const expansionResponse = await AskUserQuestion({
+  questions: [{
+    question: "是否将发现扩展为 Issue？",
+    header: "扩展选项",
+    multiSelect: true,
+    options: [
+      { label: "测试 (Test)", description: "添加更多测试用例" },
+      { label: "增强 (Enhance)", description: "功能增强建议" },
+      { label: "重构 (Refactor)", description: "代码重构建议" },
+      { label: "文档 (Doc)", description: "文档更新需求" },
+      { label: "否，直接完成", description: "不创建 Issue" }
+    ]
+  }]
+})
+
+const selectedExpansions = expansionResponse["扩展选项"]
+
+if (selectedExpansions && !selectedExpansions.includes("否，直接完成")) {
+  for (const expansion of selectedExpansions) {
+    const dimension = expansion.split(' ')[0].toLowerCase()
+    const issueSummary = `${state.task_description} - ${dimension}`
+
+    console.log(`\n创建 Issue: ${issueSummary}`)
+
+    // 调用 /issue:new 创建 issue
+    await Bash({
+      command: `/issue:new "${issueSummary}"`,
+      run_in_background: false
+    })
+  }
+}
+```
+
+### Step 4: 最终输出
+
+```javascript
+console.log(`
+═══════════════════════════════════════════════════════════
+  ✅ CCW Loop 会话完成
+═══════════════════════════════════════════════════════════
+
+  会话 ID: ${state.session_id}
+  用时: ${formatDuration(stats.duration)}
+  迭代: ${stats.iterations}
+
+  开发: ${stats.develop.completed_tasks}/${stats.develop.total_tasks} 任务完成
+  调试: ${stats.debug.iterations} 次迭代
+  验证: ${stats.validate.passed ? '通过 ✅' : '未通过 ❌'}
+
+  报告: ${sessionFolder}/summary.md
+
+═══════════════════════════════════════════════════════════
+`)
+```
+
+## State Updates
+
+```javascript
+return {
+  stateUpdates: {
+    status: 'completed',
+    completed_at: getUtc8ISOString(),
+    summary: stats
+  },
+  continue: false,
+  message: `会话 ${state.session_id} 已完成`
+}
+```
+
+## Helper Functions
+
+```javascript
+function formatDuration(ms) {
+  const seconds = Math.floor(ms / 1000)
+  const minutes = Math.floor(seconds / 60)
+  const hours = Math.floor(minutes / 60)
+
+  if (hours > 0) {
+    return `${hours}h ${minutes % 60}m`
+  } else if (minutes > 0) {
+    return `${minutes}m ${seconds % 60}s`
+  } else {
+    return `${seconds}s`
+  }
+}
+
+function generateRecommendations(stats, state) {
+  const recommendations = []
+
+  if (stats.develop.completion_rate < 100) {
+    recommendations.push('- 完成剩余开发任务')
+  }
+
+  if (!stats.validate.passed) {
+    recommendations.push('- 修复失败的测试')
+  }
+
+  if (stats.validate.coverage && stats.validate.coverage < 80) {
+    recommendations.push(`- 提高测试覆盖率 (当前: ${stats.validate.coverage}%)`)
+  }
+
+  if (stats.debug.iterations > 3 && !stats.debug.root_cause_found) {
+    recommendations.push('- 考虑代码重构以简化调试')
+  }
+
+  if (recommendations.length === 0) {
+    recommendations.push('- 考虑代码审查')
+    recommendations.push('- 更新相关文档')
+    recommendations.push('- 准备部署')
+  }
+
+  return recommendations.join('\n')
+}
+```
+
+## Error Handling
+
+| Error Type | Recovery |
+|------------|----------|
+| 报告生成失败 | 显示基本统计，跳过文件写入 |
+| Issue 创建失败 | 记录错误，继续完成 |
+
+## Next Actions
+
+- 无 (终止状态)
+- 如需继续: 使用 `ccw-loop --resume {session-id}` 重新打开会话

.claude/skills/ccw-loop/phases/actions/action-debug-with-file.md

@@ -0,0 +1,485 @@
+# Action: Debug With File
+
+假设驱动调试，记录理解演变到 understanding.md，支持 Gemini 辅助分析和假设生成。
+
+## Purpose
+
+执行假设驱动的调试流程，包括：
+- 定位错误源
+- 生成可测试假设
+- 添加 NDJSON 日志
+- 分析日志证据
+- 纠正错误理解
+- 应用修复
+
+## Preconditions
+
+- [ ] state.initialized === true
+- [ ] state.status === 'running'
+
+## Session Setup
+
+```javascript
+const getUtc8ISOString = () => new Date(Date.now() + 8 * 60 * 60 * 1000).toISOString()
+
+const sessionFolder = `.workflow/.loop/${state.session_id}`
+const debugFolder = `${sessionFolder}/debug`
+const understandingPath = `${debugFolder}/understanding.md`
+const hypothesesPath = `${debugFolder}/hypotheses.json`
+const debugLogPath = `${debugFolder}/debug.log`
+```
+
+---
+
+## Mode Detection
+
+```javascript
+// 自动检测模式
+const understandingExists = fs.existsSync(understandingPath)
+const logHasContent = fs.existsSync(debugLogPath) && fs.statSync(debugLogPath).size > 0
+
+const debugMode = logHasContent ? 'analyze' : (understandingExists ? 'continue' : 'explore')
+
+console.log(`Debug mode: ${debugMode}`)
+```
+
+---
+
+## Explore Mode (首次调试)
+
+### Step 1.1: 定位错误源
+
+```javascript
+if (debugMode === 'explore') {
+  // 询问用户 bug 描述
+  const bugInput = await AskUserQuestion({
+    questions: [{
+      question: "请描述遇到的 bug 或错误信息：",
+      header: "Bug 描述",
+      multiSelect: false,
+      options: [
+        { label: "手动输入", description: "输入错误描述或堆栈" },
+        { label: "从测试失败", description: "从验证阶段的失败测试中获取" }
+      ]
+    }]
+  })
+
+  const bugDescription = bugInput["Bug 描述"]
+
+  // 提取关键词并搜索
+  const searchResults = await Task({
+    subagent_type: 'Explore',
+    run_in_background: false,
+    prompt: `Search codebase for error patterns related to: ${bugDescription}`
+  })
+
+  // 分析搜索结果，识别受影响的位置
+  const affectedLocations = analyzeSearchResults(searchResults)
+}
+```
+
+### Step 1.2: 记录初始理解
+
+```javascript
+// 创建 understanding.md
+const initialUnderstanding = `# Understanding Document
+
+**Session ID**: ${state.session_id}
+**Bug Description**: ${bugDescription}
+**Started**: ${getUtc8ISOString()}
+
+---
+
+## Exploration Timeline
+
+### Iteration 1 - Initial Exploration (${getUtc8ISOString()})
+
+#### Current Understanding
+
+Based on bug description and initial code search:
+
+- Error pattern: ${errorPattern}
+- Affected areas: ${affectedLocations.map(l => l.file).join(', ')}
+- Initial hypothesis: ${initialThoughts}
+
+#### Evidence from Code Search
+
+${searchResults.map(r => `
+**Keyword: "${r.keyword}"**
+- Found in: ${r.files.join(', ')}
+- Key findings: ${r.insights}
+`).join('\n')}
+
+#### Next Steps
+
+- Generate testable hypotheses
+- Add instrumentation
+- Await reproduction
+
+---
+
+## Current Consolidated Understanding
+
+${initialConsolidatedUnderstanding}
+`
+
+Write(understandingPath, initialUnderstanding)
+```
+
+### Step 1.3: Gemini 辅助假设生成
+
+```bash
+ccw cli -p "
+PURPOSE: Generate debugging hypotheses for: ${bugDescription}
+Success criteria: Testable hypotheses with clear evidence criteria
+
+TASK:
+• Analyze error pattern and code search results
+• Identify 3-5 most likely root causes
+• For each hypothesis, specify:
+  - What might be wrong
+  - What evidence would confirm/reject it
+  - Where to add instrumentation
+• Rank by likelihood
+
+MODE: analysis
+
+CONTEXT: @${understandingPath} | Search results in understanding.md
+
+EXPECTED:
+- Structured hypothesis list (JSON format)
+- Each hypothesis with: id, description, testable_condition, logging_point, evidence_criteria
+- Likelihood ranking (1=most likely)
+
+CONSTRAINTS: Focus on testable conditions
+" --tool gemini --mode analysis --rule analysis-diagnose-bug-root-cause
+```
+
+### Step 1.4: 保存假设
+
+```javascript
+const hypotheses = {
+  iteration: 1,
+  timestamp: getUtc8ISOString(),
+  bug_description: bugDescription,
+  hypotheses: [
+    {
+      id: "H1",
+      description: "...",
+      testable_condition: "...",
+      logging_point: "file.ts:func:42",
+      evidence_criteria: {
+        confirm: "...",
+        reject: "..."
+      },
+      likelihood: 1,
+      status: "pending"
+    }
+    // ...
+  ],
+  gemini_insights: "...",
+  corrected_assumptions: []
+}
+
+Write(hypothesesPath, JSON.stringify(hypotheses, null, 2))
+```
+
+### Step 1.5: 添加 NDJSON 日志
+
+```javascript
+// 为每个假设添加日志点
+for (const hypothesis of hypotheses.hypotheses) {
+  const [file, func, line] = hypothesis.logging_point.split(':')
+
+  const logStatement = `console.log(JSON.stringify({
+    hid: "${hypothesis.id}",
+    ts: Date.now(),
+    func: "${func}",
+    data: { /* 相关数据 */ }
+  }))`
+
+  // 使用 Edit 工具添加日志
+  // ...
+}
+```
+
+---
+
+## Analyze Mode (有日志后)
+
+### Step 2.1: 解析调试日志
+
+```javascript
+if (debugMode === 'analyze') {
+  // 读取 NDJSON 日志
+  const logContent = Read(debugLogPath)
+  const entries = logContent.split('\n')
+    .filter(l => l.trim())
+    .map(l => JSON.parse(l))
+
+  // 按假设分组
+  const byHypothesis = groupBy(entries, 'hid')
+}
+```
+
+### Step 2.2: Gemini 辅助证据分析
+
+```bash
+ccw cli -p "
+PURPOSE: Analyze debug log evidence to validate/correct hypotheses for: ${bugDescription}
+Success criteria: Clear verdict per hypothesis + corrected understanding
+
+TASK:
+• Parse log entries by hypothesis
+• Evaluate evidence against expected criteria
+• Determine verdict: confirmed | rejected | inconclusive
+• Identify incorrect assumptions from previous understanding
+• Suggest corrections to understanding
+
+MODE: analysis
+
+CONTEXT:
+@${debugLogPath}
+@${understandingPath}
+@${hypothesesPath}
+
+EXPECTED:
+- Per-hypothesis verdict with reasoning
+- Evidence summary
+- List of incorrect assumptions with corrections
+- Updated consolidated understanding
+- Root cause if confirmed, or next investigation steps
+
+CONSTRAINTS: Evidence-based reasoning only, no speculation
+" --tool gemini --mode analysis --rule analysis-diagnose-bug-root-cause
+```
+
+### Step 2.3: 更新理解文档
+
+```javascript
+// 追加新迭代到 understanding.md
+const iteration = state.debug.iteration + 1
+
+const analysisEntry = `
+### Iteration ${iteration} - Evidence Analysis (${getUtc8ISOString()})
+
+#### Log Analysis Results
+
+${results.map(r => `
+**${r.id}**: ${r.verdict.toUpperCase()}
+- Evidence: ${JSON.stringify(r.evidence)}
+- Reasoning: ${r.reason}
+`).join('\n')}
+
+#### Corrected Understanding
+
+Previous misunderstandings identified and corrected:
+
+${corrections.map(c => `
+- ~~${c.wrong}~~ → ${c.corrected}
+  - Why wrong: ${c.reason}
+  - Evidence: ${c.evidence}
+`).join('\n')}
+
+#### New Insights
+
+${newInsights.join('\n- ')}
+
+#### Gemini Analysis
+
+${geminiAnalysis}
+
+${confirmedHypothesis ? `
+#### Root Cause Identified
+
+**${confirmedHypothesis.id}**: ${confirmedHypothesis.description}
+
+Evidence supporting this conclusion:
+${confirmedHypothesis.supportingEvidence}
+` : `
+#### Next Steps
+
+${nextSteps}
+`}
+
+---
+
+## Current Consolidated Understanding (Updated)
+
+### What We Know
+
+- ${validUnderstanding1}
+- ${validUnderstanding2}
+
+### What Was Disproven
+
+- ~~${wrongAssumption}~~ (Evidence: ${disproofEvidence})
+
+### Current Investigation Focus
+
+${currentFocus}
+
+### Remaining Questions
+
+- ${openQuestion1}
+- ${openQuestion2}
+`
+
+const existingContent = Read(understandingPath)
+Write(understandingPath, existingContent + analysisEntry)
+```
+
+### Step 2.4: 更新假设状态
+
+```javascript
+const hypothesesData = JSON.parse(Read(hypothesesPath))
+
+// 更新假设状态
+hypothesesData.hypotheses = hypothesesData.hypotheses.map(h => ({
+  ...h,
+  status: results.find(r => r.id === h.id)?.verdict || h.status,
+  evidence: results.find(r => r.id === h.id)?.evidence || h.evidence,
+  verdict_reason: results.find(r => r.id === h.id)?.reason || h.verdict_reason
+}))
+
+hypothesesData.iteration++
+hypothesesData.timestamp = getUtc8ISOString()
+
+Write(hypothesesPath, JSON.stringify(hypothesesData, null, 2))
+```
+
+---
+
+## Fix & Verification
+
+### Step 3.1: 应用修复
+
+```javascript
+if (confirmedHypothesis) {
+  console.log(`\n根因确认: ${confirmedHypothesis.description}`)
+  console.log('准备应用修复...')
+
+  // 使用 Gemini 生成修复代码
+  const fixPrompt = `
+PURPOSE: Fix the identified root cause
+Root Cause: ${confirmedHypothesis.description}
+Evidence: ${confirmedHypothesis.supportingEvidence}
+
+TASK:
+• Generate fix code
+• Ensure backward compatibility
+• Add tests if needed
+
+MODE: write
+
+CONTEXT: @${confirmedHypothesis.logging_point.split(':')[0]}
+
+EXPECTED: Fixed code + verification steps
+`
+
+  await Bash({
+    command: `ccw cli -p "${fixPrompt}" --tool gemini --mode write --rule development-debug-runtime-issues`,
+    run_in_background: false
+  })
+}
+```
+
+### Step 3.2: 记录解决方案
+
+```javascript
+const resolutionEntry = `
+### Resolution (${getUtc8ISOString()})
+
+#### Fix Applied
+
+- Modified files: ${modifiedFiles.join(', ')}
+- Fix description: ${fixDescription}
+- Root cause addressed: ${rootCause}
+
+#### Verification Results
+
+${verificationResults}
+
+#### Lessons Learned
+
+1. ${lesson1}
+2. ${lesson2}
+
+#### Key Insights for Future
+
+- ${insight1}
+- ${insight2}
+`
+
+const existingContent = Read(understandingPath)
+Write(understandingPath, existingContent + resolutionEntry)
+```
+
+### Step 3.3: 清理日志
+
+```javascript
+// 移除调试日志
+// (可选，根据用户选择)
+```
+
+---
+
+## State Updates
+
+```javascript
+return {
+  stateUpdates: {
+    debug: {
+      current_bug: bugDescription,
+      hypotheses: hypothesesData.hypotheses,
+      confirmed_hypothesis: confirmedHypothesis?.id || null,
+      iteration: hypothesesData.iteration,
+      last_analysis_at: getUtc8ISOString(),
+      understanding_updated: true
+    },
+    last_action: 'action-debug-with-file'
+  },
+  continue: true,
+  message: confirmedHypothesis
+    ? `根因确认: ${confirmedHypothesis.description}\n修复已应用，请验证`
+    : `分析完成，需要更多证据\n请复现 bug 后再次执行`
+}
+```
+
+## Error Handling
+
+| Error Type | Recovery |
+|------------|----------|
+| 空 debug.log | 提示用户复现 bug |
+| 所有假设被否定 | 使用 Gemini 生成新假设 |
+| 修复无效 | 记录失败尝试，迭代 |
+| >5 迭代 | 建议升级到 /workflow:lite-fix |
+| Gemini 不可用 | 回退到手动分析 |
+
+## Understanding Document Template
+
+参考 [templates/understanding-template.md](../../templates/understanding-template.md)
+
+## CLI Integration
+
+### 假设生成
+```bash
+ccw cli -p "PURPOSE: Generate debugging hypotheses..." --tool gemini --mode analysis --rule analysis-diagnose-bug-root-cause
+```
+
+### 证据分析
+```bash
+ccw cli -p "PURPOSE: Analyze debug log evidence..." --tool gemini --mode analysis --rule analysis-diagnose-bug-root-cause
+```
+
+### 生成修复
+```bash
+ccw cli -p "PURPOSE: Fix the identified root cause..." --tool gemini --mode write --rule development-debug-runtime-issues
+```
+
+## Next Actions (Hints)
+
+- 根因确认: `action-validate-with-file` (验证修复)
+- 需要更多证据: 等待用户复现，再次执行此动作
+- 所有假设否定: 重新执行此动作生成新假设
+- 用户选择: `action-menu` (返回菜单)

.claude/skills/ccw-loop/phases/actions/action-develop-with-file.md

@@ -0,0 +1,365 @@
+# Action: Develop With File
+
+增量开发任务执行，记录进度到 progress.md，支持 Gemini 辅助实现。
+
+## Purpose
+
+执行开发任务并记录进度，包括：
+- 分析任务需求
+- 使用 Gemini/CLI 实现代码
+- 记录代码变更
+- 更新进度文档
+
+## Preconditions
+
+- [ ] state.status === 'running'
+- [ ] state.skill_state !== null
+- [ ] state.skill_state.develop.tasks.some(t => t.status === 'pending')
+
+## Session Setup (Unified Location)
+
+```javascript
+const getUtc8ISOString = () => new Date(Date.now() + 8 * 60 * 60 * 1000).toISOString()
+
+// 统一位置: .loop/{loopId}
+const loopId = state.loop_id
+const loopFile = `.loop/${loopId}.json`
+const progressDir = `.loop/${loopId}.progress`
+const progressPath = `${progressDir}/develop.md`
+const changesLogPath = `${progressDir}/changes.log`
+```
+
+---
+
+## Execution
+
+### Step 0: Check Control Signals (CRITICAL)
+
+```javascript
+/**
+ * CRITICAL: 每个 Action 必须在开始时检查控制信号
+ * 如果 API 设置了 paused/stopped，Skill 应立即退出
+ */
+function checkControlSignals(loopId) {
+  const state = JSON.parse(Read(`.loop/${loopId}.json`))
+
+  switch (state.status) {
+    case 'paused':
+      console.log('⏸️ Loop paused by API. Exiting action.')
+      return { continue: false, reason: 'paused' }
+
+    case 'failed':
+      console.log('⏹️ Loop stopped by API. Exiting action.')
+      return { continue: false, reason: 'stopped' }
+
+    case 'running':
+      return { continue: true, reason: 'running' }
+
+    default:
+      return { continue: false, reason: 'unknown_status' }
+  }
+}
+
+// Execute check
+const control = checkControlSignals(loopId)
+if (!control.continue) {
+  return {
+    skillStateUpdates: { current_action: null },
+    continue: false,
+    message: `Action terminated: ${control.reason}`
+  }
+}
+```
+
+### Step 1: 加载任务列表
+
+```javascript
+// 读取任务列表 (从 skill_state)
+let tasks = state.skill_state?.develop?.tasks || []
+
+// 如果任务列表为空，询问用户创建
+if (tasks.length === 0) {
+  // 使用 Gemini 分析任务描述，生成任务列表
+  const analysisPrompt = `
+PURPOSE: 分析开发任务并分解为可执行步骤
+Success: 生成 3-7 个具体、可验证的子任务
+
+TASK:
+• 分析任务描述: ${state.task_description}
+• 识别关键功能点
+• 分解为独立子任务
+• 为每个子任务指定工具和模式
+
+MODE: analysis
+
+CONTEXT: @package.json @src/**/*.ts | Memory: 项目结构
+
+EXPECTED:
+JSON 格式:
+{
+  "tasks": [
+    {
+      "id": "task-001",
+      "description": "任务描述",
+      "tool": "gemini",
+      "mode": "write",
+      "files": ["src/xxx.ts"]
+    }
+  ]
+}
+`
+
+  const result = await Task({
+    subagent_type: 'cli-execution-agent',
+    run_in_background: false,
+    prompt: `Execute Gemini CLI with prompt: ${analysisPrompt}`
+  })
+
+  tasks = JSON.parse(result).tasks
+}
+
+// 找到第一个待处理任务
+const currentTask = tasks.find(t => t.status === 'pending')
+
+if (!currentTask) {
+  return {
+    skillStateUpdates: {
+      develop: { ...state.skill_state.develop, current_task: null }
+    },
+    continue: true,
+    message: '所有开发任务已完成'
+  }
+}
+```
+
+### Step 2: 执行开发任务
+
+```javascript
+console.log(`\n执行任务: ${currentTask.description}`)
+
+// 更新任务状态
+currentTask.status = 'in_progress'
+
+// 使用 Gemini 实现
+const implementPrompt = `
+PURPOSE: 实现开发任务
+Task: ${currentTask.description}
+Success criteria: 代码实现完成，测试通过
+
+TASK:
+• 分析现有代码结构
+• 实现功能代码
+• 添加必要的类型定义
+• 确保代码风格一致
+
+MODE: write
+
+CONTEXT: @${currentTask.files?.join(' @') || 'src/**/*.ts'}
+
+EXPECTED:
+- 完整的代码实现
+- 代码变更列表
+- 简要实现说明
+
+CONSTRAINTS: 遵循现有代码风格 | 不破坏现有功能
+`
+
+const implementResult = await Bash({
+  command: `ccw cli -p "${implementPrompt}" --tool gemini --mode write --rule development-implement-feature`,
+  run_in_background: false
+})
+
+// 记录代码变更
+const timestamp = getUtc8ISOString()
+const changeEntry = {
+  timestamp,
+  task_id: currentTask.id,
+  description: currentTask.description,
+  files_changed: currentTask.files || [],
+  result: 'success'
+}
+
+// 追加到 changes.log (NDJSON 格式)
+const changesContent = Read(changesLogPath) || ''
+Write(changesLogPath, changesContent + JSON.stringify(changeEntry) + '\n')
+```
+
+### Step 3: 更新进度文档
+
+```javascript
+const timestamp = getUtc8ISOString()
+const iteration = state.develop.completed_count + 1
+
+// 读取现有进度文档
+let progressContent = Read(progressPath) || ''
+
+// 如果是新文档，添加头部
+if (!progressContent) {
+  progressContent = `# Development Progress
+
+**Session ID**: ${state.session_id}
+**Task**: ${state.task_description}
+**Started**: ${timestamp}
+
+---
+
+## Progress Timeline
+
+`
+}
+
+// 追加本次进度
+const progressEntry = `
+### Iteration ${iteration} - ${currentTask.description} (${timestamp})
+
+#### Task Details
+
+- **ID**: ${currentTask.id}
+- **Tool**: ${currentTask.tool}
+- **Mode**: ${currentTask.mode}
+
+#### Implementation Summary
+
+${implementResult.summary || '实现完成'}
+
+#### Files Changed
+
+${currentTask.files?.map(f => `- \`${f}\``).join('\n') || '- No files specified'}
+
+#### Status: COMPLETED
+
+---
+
+`
+
+Write(progressPath, progressContent + progressEntry)
+
+// 更新任务状态
+currentTask.status = 'completed'
+currentTask.completed_at = timestamp
+```
+
+### Step 4: 更新任务列表文件
+
+```javascript
+// 更新 tasks.json
+const updatedTasks = tasks.map(t =>
+  t.id === currentTask.id ? currentTask : t
+)
+
+Write(tasksPath, JSON.stringify(updatedTasks, null, 2))
+```
+
+## State Updates
+
+```javascript
+return {
+  stateUpdates: {
+    develop: {
+      tasks: updatedTasks,
+      current_task_id: null,
+      completed_count: state.develop.completed_count + 1,
+      total_count: updatedTasks.length,
+      last_progress_at: getUtc8ISOString()
+    },
+    last_action: 'action-develop-with-file'
+  },
+  continue: true,
+  message: `任务完成: ${currentTask.description}\n进度: ${state.develop.completed_count + 1}/${updatedTasks.length}`
+}
+```
+
+## Error Handling
+
+| Error Type | Recovery |
+|------------|----------|
+| Gemini CLI 失败 | 提示用户手动实现，记录到 progress.md |
+| 文件写入失败 | 重试一次，失败则记录错误 |
+| 任务解析失败 | 询问用户手动输入任务 |
+
+## Progress Document Template
+
+```markdown
+# Development Progress
+
+**Session ID**: LOOP-xxx-2026-01-22
+**Task**: 实现用户认证功能
+**Started**: 2026-01-22T10:00:00+08:00
+
+---
+
+## Progress Timeline
+
+### Iteration 1 - 分析登录组件 (2026-01-22T10:05:00+08:00)
+
+#### Task Details
+
+- **ID**: task-001
+- **Tool**: gemini
+- **Mode**: analysis
+
+#### Implementation Summary
+
+分析了现有登录组件结构，识别了需要修改的文件和依赖关系。
+
+#### Files Changed
+
+- `src/components/Login.tsx`
+- `src/hooks/useAuth.ts`
+
+#### Status: COMPLETED
+
+---
+
+### Iteration 2 - 实现登录 API (2026-01-22T10:15:00+08:00)
+
+...
+
+---
+
+## Current Statistics
+
+| Metric | Value |
+|--------|-------|
+| Total Tasks | 5 |
+| Completed | 2 |
+| In Progress | 1 |
+| Pending | 2 |
+| Progress | 40% |
+
+---
+
+## Next Steps
+
+- [ ] 完成剩余任务
+- [ ] 运行测试
+- [ ] 代码审查
+```
+
+## CLI Integration
+
+### 任务分析
+```bash
+ccw cli -p "PURPOSE: 分解开发任务为子任务
+TASK: • 分析任务描述 • 识别功能点 • 生成任务列表
+MODE: analysis
+CONTEXT: @package.json @src/**/*
+EXPECTED: JSON 任务列表
+" --tool gemini --mode analysis --rule planning-breakdown-task-steps
+```
+
+### 代码实现
+```bash
+ccw cli -p "PURPOSE: 实现功能代码
+TASK: • 分析需求 • 编写代码 • 添加类型
+MODE: write
+CONTEXT: @src/xxx.ts
+EXPECTED: 完整实现
+" --tool gemini --mode write --rule development-implement-feature
+```
+
+## Next Actions (Hints)
+
+- 所有任务完成: `action-debug-with-file` (开始调试)
+- 任务失败: `action-develop-with-file` (重试或下一个任务)
+- 用户选择: `action-menu` (返回菜单)

.claude/skills/ccw-loop/phases/actions/action-init.md

@@ -0,0 +1,200 @@
+# Action: Initialize
+
+初始化 CCW Loop 会话，创建目录结构和初始状态。
+
+## Purpose
+
+- 创建会话目录结构
+- 初始化状态文件
+- 分析任务描述生成初始任务列表
+- 准备执行环境
+
+## Preconditions
+
+- [ ] state.status === 'pending'
+- [ ] state.initialized === false
+
+## Execution
+
+### Step 1: 创建目录结构
+
+```javascript
+const getUtc8ISOString = () => new Date(Date.now() + 8 * 60 * 60 * 1000).toISOString()
+
+const taskSlug = state.task_description.toLowerCase().replace(/[^a-z0-9]+/g, '-').substring(0, 30)
+const dateStr = getUtc8ISOString().substring(0, 10)
+const sessionId = `LOOP-${taskSlug}-${dateStr}`
+const sessionFolder = `.workflow/.loop/${sessionId}`
+
+Bash(`mkdir -p "${sessionFolder}/develop"`)
+Bash(`mkdir -p "${sessionFolder}/debug"`)
+Bash(`mkdir -p "${sessionFolder}/validate"`)
+
+console.log(`Session created: ${sessionId}`)
+console.log(`Location: ${sessionFolder}`)
+```
+
+### Step 2: 创建元数据文件
+
+```javascript
+const meta = {
+  session_id: sessionId,
+  task_description: state.task_description,
+  created_at: getUtc8ISOString(),
+  mode: state.mode || 'interactive'
+}
+
+Write(`${sessionFolder}/meta.json`, JSON.stringify(meta, null, 2))
+```
+
+### Step 3: 分析任务生成开发任务列表
+
+```javascript
+// 使用 Gemini 分析任务描述
+console.log('\n分析任务描述...')
+
+const analysisPrompt = `
+PURPOSE: 分析开发任务并分解为可执行步骤
+Success: 生成 3-7 个具体、可验证的子任务
+
+TASK:
+• 分析任务描述: ${state.task_description}
+• 识别关键功能点
+• 分解为独立子任务
+• 为每个子任务指定工具和模式
+
+MODE: analysis
+
+CONTEXT: @package.json @src/**/*.ts (如存在)
+
+EXPECTED:
+JSON 格式:
+{
+  "tasks": [
+    {
+      "id": "task-001",
+      "description": "任务描述",
+      "tool": "gemini",
+      "mode": "write",
+      "priority": 1
+    }
+  ],
+  "estimated_complexity": "low|medium|high",
+  "key_files": ["file1.ts", "file2.ts"]
+}
+
+CONSTRAINTS: 生成实际可执行的任务
+`
+
+const result = await Bash({
+  command: `ccw cli -p "${analysisPrompt}" --tool gemini --mode analysis --rule planning-breakdown-task-steps`,
+  run_in_background: false
+})
+
+const analysis = JSON.parse(result.stdout)
+const tasks = analysis.tasks.map((t, i) => ({
+  ...t,
+  id: t.id || `task-${String(i + 1).padStart(3, '0')}`,
+  status: 'pending',
+  created_at: getUtc8ISOString(),
+  completed_at: null,
+  files_changed: []
+}))
+
+// 保存任务列表
+Write(`${sessionFolder}/develop/tasks.json`, JSON.stringify(tasks, null, 2))
+```
+
+### Step 4: 初始化进度文档
+
+```javascript
+const progressInitial = `# Development Progress
+
+**Session ID**: ${sessionId}
+**Task**: ${state.task_description}
+**Started**: ${getUtc8ISOString()}
+**Estimated Complexity**: ${analysis.estimated_complexity}
+
+---
+
+## Task List
+
+${tasks.map((t, i) => `${i + 1}. [ ] ${t.description}`).join('\n')}
+
+## Key Files
+
+${analysis.key_files?.map(f => `- \`${f}\``).join('\n') || '- To be determined'}
+
+---
+
+## Progress Timeline
+
+`
+
+Write(`${sessionFolder}/develop/progress.md`, progressInitial)
+```
+
+### Step 5: 显示初始化结果
+
+```javascript
+console.log(`\n✅ 会话初始化完成`)
+console.log(`\n任务列表 (${tasks.length} 项):`)
+tasks.forEach((t, i) => {
+  console.log(`  ${i + 1}. ${t.description} [${t.tool}/${t.mode}]`)
+})
+console.log(`\n预估复杂度: ${analysis.estimated_complexity}`)
+console.log(`\n执行 'develop' 开始开发，或 'menu' 查看更多选项`)
+```
+
+## State Updates
+
+```javascript
+return {
+  stateUpdates: {
+    session_id: sessionId,
+    status: 'running',
+    initialized: true,
+    develop: {
+      tasks: tasks,
+      current_task_id: null,
+      completed_count: 0,
+      total_count: tasks.length,
+      last_progress_at: null
+    },
+    debug: {
+      current_bug: null,
+      hypotheses: [],
+      confirmed_hypothesis: null,
+      iteration: 0,
+      last_analysis_at: null,
+      understanding_updated: false
+    },
+    validate: {
+      test_results: [],
+      coverage: null,
+      passed: false,
+      failed_tests: [],
+      last_run_at: null
+    },
+    context: {
+      estimated_complexity: analysis.estimated_complexity,
+      key_files: analysis.key_files
+    }
+  },
+  continue: true,
+  message: `会话 ${sessionId} 已初始化\n${tasks.length} 个开发任务待执行`
+}
+```
+
+## Error Handling
+
+| Error Type | Recovery |
+|------------|----------|
+| 目录创建失败 | 检查权限，重试 |
+| Gemini 分析失败 | 提示用户手动输入任务 |
+| 任务解析失败 | 使用默认任务列表 |
+
+## Next Actions
+
+- 成功: `action-menu` (显示操作菜单) 或 `action-develop-with-file` (直接开始开发)
+- 失败: 报错退出

.claude/skills/ccw-loop/phases/actions/action-menu.md

@@ -0,0 +1,192 @@
+# Action: Menu
+
+显示交互式操作菜单，让用户选择下一步操作。
+
+## Purpose
+
+- 显示当前状态摘要
+- 提供操作选项
+- 接收用户选择
+- 返回下一个动作
+
+## Preconditions
+
+- [ ] state.initialized === true
+- [ ] state.status === 'running'
+
+## Execution
+
+### Step 1: 生成状态摘要
+
+```javascript
+const getUtc8ISOString = () => new Date(Date.now() + 8 * 60 * 60 * 1000).toISOString()
+
+// 开发进度
+const developProgress = state.develop.total_count > 0
+  ? `${state.develop.completed_count}/${state.develop.total_count} (${(state.develop.completed_count / state.develop.total_count * 100).toFixed(0)}%)`
+  : '未开始'
+
+// 调试状态
+const debugStatus = state.debug.confirmed_hypothesis
+  ? `✅ 已确认根因`
+  : state.debug.iteration > 0
+    ? `🔍 迭代 ${state.debug.iteration}`
+    : '未开始'
+
+// 验证状态
+const validateStatus = state.validate.passed
+  ? `✅ 通过`
+  : state.validate.test_results.length > 0
+    ? `❌ ${state.validate.failed_tests.length} 个失败`
+    : '未运行'
+
+const statusSummary = `
+═══════════════════════════════════════════════════════════
+  CCW Loop - ${state.session_id}
+═══════════════════════════════════════════════════════════
+
+  任务: ${state.task_description}
+  迭代: ${state.iteration_count}
+
+  ┌─────────────────────────────────────────────────────┐
+  │  开发 (Develop)  │  ${developProgress.padEnd(20)}      │
+  │  调试 (Debug)    │  ${debugStatus.padEnd(20)}      │
+  │  验证 (Validate) │  ${validateStatus.padEnd(20)}      │
+  └─────────────────────────────────────────────────────┘
+
+═══════════════════════════════════════════════════════════
+`
+
+console.log(statusSummary)
+```
+
+### Step 2: 显示操作选项
+
+```javascript
+const options = [
+  {
+    label: "📝 继续开发 (Develop)",
+    description: state.develop.completed_count < state.develop.total_count
+      ? `执行下一个开发任务`
+      : "所有任务已完成，可添加新任务",
+    action: "action-develop-with-file"
+  },
+  {
+    label: "🔍 开始调试 (Debug)",
+    description: state.debug.iteration > 0
+      ? "继续假设驱动调试"
+      : "开始新的调试会话",
+    action: "action-debug-with-file"
+  },
+  {
+    label: "✅ 运行验证 (Validate)",
+    description: "运行测试并检查覆盖率",
+    action: "action-validate-with-file"
+  },
+  {
+    label: "📊 查看详情 (Status)",
+    description: "查看详细进度和文件",
+    action: "action-status"
+  },
+  {
+    label: "🏁 完成循环 (Complete)",
+    description: "结束当前循环",
+    action: "action-complete"
+  },
+  {
+    label: "🚪 退出 (Exit)",
+    description: "保存状态并退出",
+    action: "exit"
+  }
+]
+
+const response = await AskUserQuestion({
+  questions: [{
+    question: "选择下一步操作：",
+    header: "操作",
+    multiSelect: false,
+    options: options.map(o => ({
+      label: o.label,
+      description: o.description
+    }))
+  }]
+})
+
+const selectedLabel = response["操作"]
+const selectedOption = options.find(o => o.label === selectedLabel)
+const nextAction = selectedOption?.action || 'action-menu'
+```
+
+### Step 3: 处理特殊选项
+
+```javascript
+if (nextAction === 'exit') {
+  console.log('\n保存状态并退出...')
+  return {
+    stateUpdates: {
+      status: 'user_exit'
+    },
+    continue: false,
+    message: '会话已保存，使用 --resume 可继续'
+  }
+}
+
+if (nextAction === 'action-status') {
+  // 显示详细状态
+  const sessionFolder = `.workflow/.loop/${state.session_id}`
+
+  console.log('\n=== 开发进度 ===')
+  const progress = Read(`${sessionFolder}/develop/progress.md`)
+  console.log(progress?.substring(0, 500) + '...')
+
+  console.log('\n=== 调试状态 ===')
+  if (state.debug.hypotheses.length > 0) {
+    state.debug.hypotheses.forEach(h => {
+      console.log(`  ${h.id}: ${h.status} - ${h.description.substring(0, 50)}...`)
+    })
+  } else {
+    console.log('  尚未开始调试')
+  }
+
+  console.log('\n=== 验证结果 ===')
+  if (state.validate.test_results.length > 0) {
+    const latest = state.validate.test_results[state.validate.test_results.length - 1]
+    console.log(`  最近运行: ${latest.timestamp}`)
+    console.log(`  通过率: ${latest.summary.pass_rate}%`)
+  } else {
+    console.log('  尚未运行验证')
+  }
+
+  // 返回菜单
+  return {
+    stateUpdates: {},
+    continue: true,
+    nextAction: 'action-menu',
+    message: ''
+  }
+}
+```
+
+## State Updates
+
+```javascript
+return {
+  stateUpdates: {
+    // 不更新状态，仅返回下一个动作
+  },
+  continue: true,
+  nextAction: nextAction,
+  message: `执行: ${selectedOption?.label || nextAction}`
+}
+```
+
+## Error Handling
+
+| Error Type | Recovery |
+|------------|----------|
+| 用户取消 | 返回菜单 |
+| 无效选择 | 重新显示菜单 |
+
+## Next Actions
+
+根据用户选择动态决定下一个动作。

.claude/skills/ccw-loop/phases/actions/action-validate-with-file.md

@@ -0,0 +1,307 @@
+# Action: Validate With File
+
+运行测试并验证实现，记录结果到 validation.md，支持 Gemini 辅助分析测试覆盖率和质量。
+
+## Purpose
+
+执行测试验证流程，包括:
+- 运行单元测试
+- 运行集成测试
+- 检查代码覆盖率
+- 生成验证报告
+- 分析失败原因
+
+## Preconditions
+
+- [ ] state.initialized === true
+- [ ] state.status === 'running'
+- [ ] state.develop.completed_count > 0 || state.debug.confirmed_hypothesis !== null
+
+## Session Setup
+
+```javascript
+const getUtc8ISOString = () => new Date(Date.now() + 8 * 60 * 60 * 1000).toISOString()
+
+const sessionFolder = `.workflow/.loop/${state.session_id}`
+const validateFolder = `${sessionFolder}/validate`
+const validationPath = `${validateFolder}/validation.md`
+const testResultsPath = `${validateFolder}/test-results.json`
+const coveragePath = `${validateFolder}/coverage.json`
+```
+
+---
+
+## Execution
+
+### Step 1: 运行测试
+
+```javascript
+console.log('\n运行测试...')
+
+// 检测测试框架
+const packageJson = JSON.parse(Read('package.json'))
+const testScript = packageJson.scripts?.test || 'npm test'
+
+// 运行测试并捕获输出
+const testResult = await Bash({
+  command: testScript,
+  timeout: 300000  // 5分钟
+})
+
+// 解析测试输出
+const testResults = parseTestOutput(testResult.stdout)
+```
+
+### Step 2: 检查覆盖率
+
+```javascript
+// 运行覆盖率检查
+let coverageData = null
+
+if (packageJson.scripts?.['test:coverage']) {
+  const coverageResult = await Bash({
+    command: 'npm run test:coverage',
+    timeout: 300000
+  })
+
+  // 解析覆盖率报告
+  coverageData = parseCoverageReport(coverageResult.stdout)
+
+  Write(coveragePath, JSON.stringify(coverageData, null, 2))
+}
+```
+
+### Step 3: Gemini 辅助分析
+
+```bash
+ccw cli -p "
+PURPOSE: Analyze test results and coverage
+Success criteria: Identify quality issues and suggest improvements
+
+TASK:
+• Analyze test execution results
+• Review code coverage metrics
+• Identify missing test cases
+• Suggest quality improvements
+• Verify requirements coverage
+
+MODE: analysis
+
+CONTEXT:
+@${testResultsPath}
+@${coveragePath}
+@${sessionFolder}/develop/progress.md
+
+EXPECTED:
+- Quality assessment report
+- Failed tests analysis
+- Coverage gaps identification
+- Improvement recommendations
+- Pass/Fail decision with rationale
+
+CONSTRAINTS: Evidence-based quality assessment
+" --tool gemini --mode analysis --rule analysis-review-code-quality
+```
+
+### Step 4: 生成验证报告
+
+```javascript
+const timestamp = getUtc8ISOString()
+const iteration = (state.validate.test_results?.length || 0) + 1
+
+const validationReport = `# Validation Report
+
+**Session ID**: ${state.session_id}
+**Task**: ${state.task_description}
+**Validated**: ${timestamp}
+
+---
+
+## Iteration ${iteration} - Validation Run
+
+### Test Execution Summary
+
+| Metric | Value |
+|--------|-------|
+| Total Tests | ${testResults.total} |
+| Passed | ${testResults.passed} |
+| Failed | ${testResults.failed} |
+| Skipped | ${testResults.skipped} |
+| Duration | ${testResults.duration_ms}ms |
+| **Pass Rate** | **${(testResults.passed / testResults.total * 100).toFixed(1)}%** |
+
+### Coverage Report
+
+${coverageData ? `
+| File | Statements | Branches | Functions | Lines |
+|------|------------|----------|-----------|-------|
+${coverageData.files.map(f => `| ${f.path} | ${f.statements}% | ${f.branches}% | ${f.functions}% | ${f.lines}% |`).join('\n')}
+
+**Overall Coverage**: ${coverageData.overall.statements}%
+` : '_No coverage data available_'}
+
+### Failed Tests
+
+${testResults.failed > 0 ? `
+${testResults.failures.map(f => `
+#### ${f.test_name}
+
+- **Suite**: ${f.suite}
+- **Error**: ${f.error_message}
+- **Stack**:
+\`\`\`
+${f.stack_trace}
+\`\`\`
+`).join('\n')}
+` : '_All tests passed_'}
+
+### Gemini Quality Analysis
+
+${geminiAnalysis}
+
+### Recommendations
+
+${recommendations.map(r => `- ${r}`).join('\n')}
+
+---
+
+## Validation Decision
+
+**Result**: ${testResults.passed === testResults.total ? '✅ PASS' : '❌ FAIL'}
+
+**Rationale**: ${validationDecision}
+
+${testResults.passed !== testResults.total ? `
+### Next Actions
+
+1. Review failed tests
+2. Debug failures using action-debug-with-file
+3. Fix issues and re-run validation
+` : `
+### Next Actions
+
+1. Consider code review
+2. Prepare for deployment
+3. Update documentation
+`}
+`
+
+// 写入验证报告
+Write(validationPath, validationReport)
+```
+
+### Step 5: 保存测试结果
+
+```javascript
+const testResultsData = {
+  iteration,
+  timestamp,
+  summary: {
+    total: testResults.total,
+    passed: testResults.passed,
+    failed: testResults.failed,
+    skipped: testResults.skipped,
+    pass_rate: (testResults.passed / testResults.total * 100).toFixed(1),
+    duration_ms: testResults.duration_ms
+  },
+  tests: testResults.tests,
+  failures: testResults.failures,
+  coverage: coverageData?.overall || null
+}
+
+Write(testResultsPath, JSON.stringify(testResultsData, null, 2))
+```
+
+---
+
+## State Updates
+
+```javascript
+const validationPassed = testResults.failed === 0 && testResults.passed > 0
+
+return {
+  stateUpdates: {
+    validate: {
+      test_results: [...(state.validate.test_results || []), testResultsData],
+      coverage: coverageData?.overall.statements || null,
+      passed: validationPassed,
+      failed_tests: testResults.failures.map(f => f.test_name),
+      last_run_at: getUtc8ISOString()
+    },
+    last_action: 'action-validate-with-file'
+  },
+  continue: true,
+  message: validationPassed
+    ? `验证通过 ✅\n测试: ${testResults.passed}/${testResults.total}\n覆盖率: ${coverageData?.overall.statements || 'N/A'}%`
+    : `验证失败 ❌\n失败: ${testResults.failed}/${testResults.total}\n建议进入调试模式`
+}
+```
+
+## Test Output Parsers
+
+### Jest/Vitest Parser
+
+```javascript
+function parseJestOutput(stdout) {
+  const testPattern = /Tests:\s+(\d+) passed.*?(\d+) failed.*?(\d+) total/
+  const match = stdout.match(testPattern)
+
+  return {
+    total: parseInt(match[3]),
+    passed: parseInt(match[1]),
+    failed: parseInt(match[2]),
+    // ... parse individual test results
+  }
+}
+```
+
+### Pytest Parser
+
+```javascript
+function parsePytestOutput(stdout) {
+  const summaryPattern = /(\d+) passed.*?(\d+) failed.*?(\d+) error/
+  // ... implementation
+}
+```
+
+## Error Handling
+
+| Error Type | Recovery |
+|------------|----------|
+| Tests don't run | 检查测试脚本配置，提示用户 |
+| All tests fail | 建议进入 debug 模式 |
+| Coverage tool missing | 跳过覆盖率检查，仅运行测试 |
+| Timeout | 增加超时时间或拆分测试 |
+
+## Validation Report Template
+
+参考 [templates/validation-template.md](../../templates/validation-template.md)
+
+## CLI Integration
+
+### 质量分析
+```bash
+ccw cli -p "PURPOSE: Analyze test results and coverage...
+TASK: • Review results • Identify gaps • Suggest improvements
+MODE: analysis
+CONTEXT: @test-results.json @coverage.json
+EXPECTED: Quality assessment
+" --tool gemini --mode analysis --rule analysis-review-code-quality
+```
+
+### 测试生成 (如覆盖率低)
+```bash
+ccw cli -p "PURPOSE: Generate missing test cases...
+TASK: • Analyze uncovered code • Write tests
+MODE: write
+CONTEXT: @coverage.json @src/**/*
+EXPECTED: Test code
+" --tool gemini --mode write --rule development-generate-tests
+```
+
+## Next Actions (Hints)
+
+- 验证通过: `action-complete` (完成循环)
+- 验证失败: `action-debug-with-file` (调试失败测试)
+- 覆盖率低: `action-develop-with-file` (添加测试)
+- 用户选择: `action-menu` (返回菜单)

.claude/skills/ccw-loop/phases/orchestrator.md

@@ -0,0 +1,486 @@
+# Orchestrator
+
+根据当前状态选择并执行下一个动作，实现无状态循环工作流。与 API (loop-v2-routes.ts) 协作实现控制平面/执行平面分离。
+
+## Role
+
+检查控制信号 → 读取文件状态 → 选择动作 → 执行 → 更新文件 → 循环，直到完成或被外部暂停/停止。
+
+## State Management (Unified Location)
+
+### 读取状态
+
+```javascript
+const getUtc8ISOString = () => new Date(Date.now() + 8 * 60 * 60 * 1000).toISOString()
+
+/**
+ * 读取循环状态 (统一位置)
+ * @param loopId - Loop ID (e.g., "loop-v2-20260122-abc123")
+ */
+function readLoopState(loopId) {
+  const stateFile = `.loop/${loopId}.json`
+
+  if (!fs.existsSync(stateFile)) {
+    return null
+  }
+
+  const state = JSON.parse(Read(stateFile))
+  return state
+}
+```
+
+### 更新状态
+
+```javascript
+/**
+ * 更新循环状态 (只更新 skill_state 部分，不修改 API 字段)
+ * @param loopId - Loop ID
+ * @param updates - 更新内容 (skill_state 字段)
+ */
+function updateLoopState(loopId, updates) {
+  const stateFile = `.loop/${loopId}.json`
+  const currentState = readLoopState(loopId)
+
+  if (!currentState) {
+    throw new Error(`Loop state not found: ${loopId}`)
+  }
+
+  // 只更新 skill_state 和 updated_at
+  const newState = {
+    ...currentState,
+    updated_at: getUtc8ISOString(),
+    skill_state: {
+      ...currentState.skill_state,
+      ...updates
+    }
+  }
+
+  Write(stateFile, JSON.stringify(newState, null, 2))
+  return newState
+}
+```
+
+### 创建新循环状态 (直接调用时)
+
+```javascript
+/**
+ * 创建新的循环状态 (仅在直接调用时使用，API 触发时状态已存在)
+ */
+function createLoopState(loopId, taskDescription) {
+  const stateFile = `.loop/${loopId}.json`
+  const now = getUtc8ISOString()
+
+  const state = {
+    // API 兼容字段
+    loop_id: loopId,
+    title: taskDescription.substring(0, 100),
+    description: taskDescription,
+    max_iterations: 10,
+    status: 'running',  // 直接调用时设为 running
+    current_iteration: 0,
+    created_at: now,
+    updated_at: now,
+
+    // Skill 扩展字段
+    skill_state: null  // 由 action-init 初始化
+  }
+
+  // 确保目录存在
+  Bash(`mkdir -p ".loop"`)
+  Bash(`mkdir -p ".loop/${loopId}.progress"`)
+
+  Write(stateFile, JSON.stringify(state, null, 2))
+  return state
+}
+```
+
+## Control Signal Checking
+
+```javascript
+/**
+ * 检查 API 控制信号
+ * 必须在每个 Action 开始前调用
+ * @returns { continue: boolean, reason: string }
+ */
+function checkControlSignals(loopId) {
+  const state = readLoopState(loopId)
+
+  if (!state) {
+    return { continue: false, reason: 'state_not_found' }
+  }
+
+  switch (state.status) {
+    case 'paused':
+      // API 暂停了循环，Skill 应退出等待 resume
+      console.log(`⏸️ Loop paused by API. Waiting for resume...`)
+      return { continue: false, reason: 'paused' }
+
+    case 'failed':
+      // API 停止了循环 (用户手动停止)
+      console.log(`⏹️ Loop stopped by API.`)
+      return { continue: false, reason: 'stopped' }
+
+    case 'completed':
+      // 已完成
+      console.log(`✅ Loop already completed.`)
+      return { continue: false, reason: 'completed' }
+
+    case 'created':
+      // API 创建但未启动 (不应该走到这里)
+      console.log(`⚠️ Loop not started by API.`)
+      return { continue: false, reason: 'not_started' }
+
+    case 'running':
+      // 正常继续
+      return { continue: true, reason: 'running' }
+
+    default:
+      console.log(`⚠️ Unknown status: ${state.status}`)
+      return { continue: false, reason: 'unknown_status' }
+  }
+}
+```
+
+## Decision Logic
+
+```javascript
+/**
+ * 选择下一个 Action (基于 skill_state)
+ */
+function selectNextAction(state, mode = 'interactive') {
+  const skillState = state.skill_state
+
+  // 1. 终止条件检查 (API status)
+  if (state.status === 'completed') return null
+  if (state.status === 'failed') return null
+  if (state.current_iteration >= state.max_iterations) {
+    console.warn(`已达到最大迭代次数 (${state.max_iterations})`)
+    return 'action-complete'
+  }
+
+  // 2. 初始化检查
+  if (!skillState || !skillState.current_action) {
+    return 'action-init'
+  }
+
+  // 3. 模式判断
+  if (mode === 'interactive') {
+    return 'action-menu'  // 显示菜单让用户选择
+  }
+
+  // 4. 自动模式：基于状态自动选择
+  if (mode === 'auto') {
+    // 按优先级：develop → debug → validate
+
+    // 如果有待开发任务
+    const hasPendingDevelop = skillState.develop?.tasks?.some(t => t.status === 'pending')
+    if (hasPendingDevelop) {
+      return 'action-develop-with-file'
+    }
+
+    // 如果开发完成但未调试
+    if (skillState.last_action === 'action-develop-with-file') {
+      const needsDebug = skillState.develop?.completed < skillState.develop?.total
+      if (needsDebug) {
+        return 'action-debug-with-file'
+      }
+    }
+
+    // 如果调试完成但未验证
+    if (skillState.last_action === 'action-debug-with-file' ||
+        skillState.debug?.confirmed_hypothesis) {
+      return 'action-validate-with-file'
+    }
+
+    // 如果验证失败，回到开发
+    if (skillState.last_action === 'action-validate-with-file') {
+      if (!skillState.validate?.passed) {
+        return 'action-develop-with-file'
+      }
+    }
+
+    // 全部通过，完成
+    if (skillState.validate?.passed && !hasPendingDevelop) {
+      return 'action-complete'
+    }
+
+    // 默认：开发
+    return 'action-develop-with-file'
+  }
+
+  // 5. 默认完成
+  return 'action-complete'
+}
+```
+
+## Execution Loop
+
+```javascript
+/**
+ * 运行编排器
+ * @param options.loopId - 现有 Loop ID (API 触发时)
+ * @param options.task - 任务描述 (直接调用时)
+ * @param options.mode - 'interactive' | 'auto'
+ */
+async function runOrchestrator(options = {}) {
+  const { loopId: existingLoopId, task, mode = 'interactive' } = options
+
+  console.log('=== CCW Loop Orchestrator Started ===')
+
+  // 1. 确定 loopId
+  let loopId
+  let state
+
+  if (existingLoopId) {
+    // API 触发：使用现有 loopId
+    loopId = existingLoopId
+    state = readLoopState(loopId)
+
+    if (!state) {
+      console.error(`Loop not found: ${loopId}`)
+      return { status: 'error', message: 'Loop not found' }
+    }
+
+    console.log(`Resuming loop: ${loopId}`)
+    console.log(`Status: ${state.status}`)
+
+  } else if (task) {
+    // 直接调用：创建新 loopId
+    const timestamp = getUtc8ISOString().replace(/[-:]/g, '').split('.')[0]
+    const random = Math.random().toString(36).substring(2, 10)
+    loopId = `loop-v2-${timestamp}-${random}`
+
+    console.log(`Creating new loop: ${loopId}`)
+    console.log(`Task: ${task}`)
+
+    state = createLoopState(loopId, task)
+
+  } else {
+    console.error('Either --loop-id or task description is required')
+    return { status: 'error', message: 'Missing loopId or task' }
+  }
+
+  const progressDir = `.loop/${loopId}.progress`
+
+  // 2. 主循环
+  let iteration = state.current_iteration || 0
+
+  while (iteration < state.max_iterations) {
+    iteration++
+
+    // ========================================
+    // CRITICAL: Check control signals first
+    // ========================================
+    const control = checkControlSignals(loopId)
+    if (!control.continue) {
+      console.log(`\n🛑 Loop terminated: ${control.reason}`)
+      break
+    }
+
+    // 重新读取状态 (可能被 API 更新)
+    state = readLoopState(loopId)
+
+    console.log(`\n[Iteration ${iteration}] Status: ${state.status}`)
+
+    // 选择下一个动作
+    const actionId = selectNextAction(state, mode)
+
+    if (!actionId) {
+      console.log('No action selected, terminating.')
+      break
+    }
+
+    console.log(`[Iteration ${iteration}] Executing: ${actionId}`)
+
+    // 更新 current_iteration
+    state = {
+      ...state,
+      current_iteration: iteration,
+      updated_at: getUtc8ISOString()
+    }
+    Write(`.loop/${loopId}.json`, JSON.stringify(state, null, 2))
+
+    // 执行动作
+    try {
+      const actionPromptFile = `.claude/skills/ccw-loop/phases/actions/${actionId}.md`
+
+      if (!fs.existsSync(actionPromptFile)) {
+        console.error(`Action file not found: ${actionPromptFile}`)
+        continue
+      }
+
+      const actionPrompt = Read(actionPromptFile)
+
+      // 构建 Agent 提示
+      const agentPrompt = `
+[LOOP CONTEXT]
+Loop ID: ${loopId}
+State File: .loop/${loopId}.json
+Progress Dir: ${progressDir}
+
+[CURRENT STATE]
+${JSON.stringify(state, null, 2)}
+
+[ACTION INSTRUCTIONS]
+${actionPrompt}
+
+[TASK]
+You are executing ${actionId} for loop: ${state.title || state.description}
+
+[CONTROL SIGNALS]
+Before executing, check if status is still 'running'.
+If status is 'paused' or 'failed', exit gracefully.
+
+[RETURN]
+Return JSON with:
+- skillStateUpdates: Object with skill_state fields to update
+- continue: Boolean indicating if loop should continue
+- message: String with user message
+`
+
+      const result = await Task({
+        subagent_type: 'universal-executor',
+        run_in_background: false,
+        description: `Execute ${actionId}`,
+        prompt: agentPrompt
+      })
+
+      // 解析结果
+      const actionResult = JSON.parse(result)
+
+      // 更新状态 (只更新 skill_state)
+      updateLoopState(loopId, {
+        current_action: null,
+        last_action: actionId,
+        completed_actions: [
+          ...(state.skill_state?.completed_actions || []),
+          actionId
+        ],
+        ...actionResult.skillStateUpdates
+      })
+
+      // 显示消息
+      if (actionResult.message) {
+        console.log(`\n${actionResult.message}`)
+      }
+
+      // 检查是否继续
+      if (actionResult.continue === false) {
+        console.log('Action requested termination.')
+        break
+      }
+
+    } catch (error) {
+      console.error(`Error executing ${actionId}: ${error.message}`)
+
+      // 错误处理
+      updateLoopState(loopId, {
+        current_action: null,
+        errors: [
+          ...(state.skill_state?.errors || []),
+          {
+            action: actionId,
+            message: error.message,
+            timestamp: getUtc8ISOString()
+          }
+        ]
+      })
+    }
+  }
+
+  if (iteration >= state.max_iterations) {
+    console.log(`\n⚠️ Reached maximum iterations (${state.max_iterations})`)
+    console.log('Consider breaking down the task or taking a break.')
+  }
+
+  console.log('\n=== CCW Loop Orchestrator Finished ===')
+
+  // 返回最终状态
+  const finalState = readLoopState(loopId)
+  return {
+    status: finalState.status,
+    loop_id: loopId,
+    iterations: iteration,
+    final_state: finalState
+  }
+}
+```
+
+## Action Catalog
+
+| Action | Purpose | Preconditions | Effects |
+|--------|---------|---------------|---------|
+| [action-init](actions/action-init.md) | 初始化会话 | status=pending | initialized=true |
+| [action-menu](actions/action-menu.md) | 显示操作菜单 | initialized=true | 用户选择下一动作 |
+| [action-develop-with-file](actions/action-develop-with-file.md) | 开发任务 | initialized=true | 更新 progress.md |
+| [action-debug-with-file](actions/action-debug-with-file.md) | 假设调试 | initialized=true | 更新 understanding.md |
+| [action-validate-with-file](actions/action-validate-with-file.md) | 测试验证 | initialized=true | 更新 validation.md |
+| [action-complete](actions/action-complete.md) | 完成循环 | validation_passed=true | status=completed |
+
+## Termination Conditions
+
+1. **API 暂停**: `state.status === 'paused'` (Skill 退出，等待 resume)
+2. **API 停止**: `state.status === 'failed'` (Skill 终止)
+3. **任务完成**: `state.status === 'completed'`
+4. **迭代限制**: `state.current_iteration >= state.max_iterations`
+5. **Action 请求终止**: `actionResult.continue === false`
+
+## Error Recovery
+
+| Error Type | Recovery Strategy |
+|------------|-------------------|
+| 动作执行失败 | 记录错误，增加 error_count，继续下一动作 |
+| 状态文件损坏 | 从其他文件重建状态 (progress.md, understanding.md 等) |
+| 用户中止 | 保存当前状态，允许 --resume 恢复 |
+| CLI 工具失败 | 回退到手动分析模式 |
+
+## Mode Strategies
+
+### Interactive Mode (默认)
+
+每次显示菜单，让用户选择动作：
+
+```
+当前状态: 开发中
+可用操作:
+  1. 继续开发 (develop)
+  2. 开始调试 (debug)
+  3. 运行验证 (validate)
+  4. 查看进度 (status)
+  5. 退出 (exit)
+
+请选择:
+```
+
+### Auto Mode (自动循环)
+
+按预设流程自动执行：
+
+```
+Develop → Debug → Validate →
+  ↓ (如验证失败)
+Develop (修复) → Debug → Validate → 完成
+```
+
+## State Machine (API Status)
+
+```mermaid
+stateDiagram-v2
+    [*] --> created: API creates loop
+    created --> running: API /start → Trigger Skill
+    running --> paused: API /pause → Set status
+    running --> completed: action-complete
+    running --> failed: API /stop OR error
+    paused --> running: API /resume → Re-trigger Skill
+    completed --> [*]
+    failed --> [*]
+
+    note right of paused
+      Skill checks status before each action
+      If paused, Skill exits gracefully
+    end note
+
+    note right of running
+      Skill executes: init → develop → debug → validate
+    end note
+```

.claude/skills/ccw-loop/phases/state-schema.md

@@ -0,0 +1,474 @@
+# State Schema
+
+CCW Loop 的状态结构定义（统一版本）。
+
+## 状态文件
+
+**位置**: `.loop/{loopId}.json` (统一位置，API + Skill 共享)
+
+**旧版本位置** (仅向后兼容): `.workflow/.loop/{session-id}/state.json`
+
+## 结构定义
+
+### 统一状态接口 (Unified Loop State)
+
+```typescript
+/**
+ * Unified Loop State - API 和 Skill 共享的状态结构
+ * API (loop-v2-routes.ts) 拥有状态的主控权
+ * Skill (ccw-loop) 读取和更新此状态
+ */
+interface LoopState {
+  // =====================================================
+  // API FIELDS (from loop-v2-routes.ts)
+  // 这些字段由 API 管理，Skill 只读
+  // =====================================================
+
+  loop_id: string                // Loop ID, e.g., "loop-v2-20260122-abc123"
+  title: string                  // Loop 标题
+  description: string            // Loop 描述
+  max_iterations: number         // 最大迭代次数
+  status: 'created' | 'running' | 'paused' | 'completed' | 'failed'
+  current_iteration: number      // 当前迭代次数
+  created_at: string             // 创建时间 (ISO8601)
+  updated_at: string             // 最后更新时间 (ISO8601)
+  completed_at?: string          // 完成时间 (ISO8601)
+  failure_reason?: string        // 失败原因
+
+  // =====================================================
+  // SKILL EXTENSION FIELDS
+  // 这些字段由 Skill 管理，API 只读
+  // =====================================================
+
+  skill_state?: {
+    // 当前执行动作
+    current_action: 'init' | 'develop' | 'debug' | 'validate' | 'complete' | null
+    last_action: string | null
+    completed_actions: string[]
+    mode: 'interactive' | 'auto'
+
+    // === 开发阶段 ===
+    develop: {
+      total: number
+      completed: number
+      current_task?: string
+      tasks: DevelopTask[]
+      last_progress_at: string | null
+    }
+
+    // === 调试阶段 ===
+    debug: {
+      active_bug?: string
+      hypotheses_count: number
+      hypotheses: Hypothesis[]
+      confirmed_hypothesis: string | null
+      iteration: number
+      last_analysis_at: string | null
+    }
+
+    // === 验证阶段 ===
+    validate: {
+      pass_rate: number           // 测试通过率 (0-100)
+      coverage: number            // 覆盖率 (0-100)
+      test_results: TestResult[]
+      passed: boolean
+      failed_tests: string[]
+      last_run_at: string | null
+    }
+
+    // === 错误追踪 ===
+    errors: Array<{
+      action: string
+      message: string
+      timestamp: string
+    }>
+  }
+}
+
+interface DevelopTask {
+  id: string
+  description: string
+  tool: 'gemini' | 'qwen' | 'codex' | 'bash'
+  mode: 'analysis' | 'write'
+  status: 'pending' | 'in_progress' | 'completed' | 'failed'
+  files_changed: string[]
+  created_at: string
+  completed_at: string | null
+}
+
+interface Hypothesis {
+  id: string                    // H1, H2, ...
+  description: string
+  testable_condition: string
+  logging_point: string
+  evidence_criteria: {
+    confirm: string
+    reject: string
+  }
+  likelihood: number            // 1 = 最可能
+  status: 'pending' | 'confirmed' | 'rejected' | 'inconclusive'
+  evidence: Record<string, any> | null
+  verdict_reason: string | null
+}
+
+interface TestResult {
+  test_name: string
+  suite: string
+  status: 'passed' | 'failed' | 'skipped'
+  duration_ms: number
+  error_message: string | null
+  stack_trace: string | null
+}
+```
+
+## 初始状态
+
+### 由 API 创建时 (Dashboard 触发)
+
+```json
+{
+  "loop_id": "loop-v2-20260122-abc123",
+  "title": "Implement user authentication",
+  "description": "Add login/logout functionality",
+  "max_iterations": 10,
+  "status": "created",
+  "current_iteration": 0,
+  "created_at": "2026-01-22T10:00:00+08:00",
+  "updated_at": "2026-01-22T10:00:00+08:00"
+}
+```
+
+### 由 Skill 初始化后 (action-init)
+
+```json
+{
+  "loop_id": "loop-v2-20260122-abc123",
+  "title": "Implement user authentication",
+  "description": "Add login/logout functionality",
+  "max_iterations": 10,
+  "status": "running",
+  "current_iteration": 0,
+  "created_at": "2026-01-22T10:00:00+08:00",
+  "updated_at": "2026-01-22T10:00:05+08:00",
+
+  "skill_state": {
+    "current_action": "init",
+    "last_action": null,
+    "completed_actions": [],
+    "mode": "auto",
+
+    "develop": {
+      "total": 3,
+      "completed": 0,
+      "current_task": null,
+      "tasks": [
+        { "id": "task-001", "description": "Create auth component", "status": "pending" }
+      ],
+      "last_progress_at": null
+    },
+
+    "debug": {
+      "active_bug": null,
+      "hypotheses_count": 0,
+      "hypotheses": [],
+      "confirmed_hypothesis": null,
+      "iteration": 0,
+      "last_analysis_at": null
+    },
+
+    "validate": {
+      "pass_rate": 0,
+      "coverage": 0,
+      "test_results": [],
+      "passed": false,
+      "failed_tests": [],
+      "last_run_at": null
+    },
+
+    "errors": []
+  }
+}
+```
+
+## 控制信号检查 (Control Signals)
+
+Skill 在每个 Action 开始前必须检查控制信号：
+
+```javascript
+/**
+ * 检查 API 控制信号
+ * @returns { continue: boolean, action: 'pause_exit' | 'stop_exit' | 'continue' }
+ */
+function checkControlSignals(loopId) {
+  const state = JSON.parse(Read(`.loop/${loopId}.json`))
+
+  switch (state.status) {
+    case 'paused':
+      // API 暂停了循环，Skill 应退出等待 resume
+      return { continue: false, action: 'pause_exit' }
+
+    case 'failed':
+      // API 停止了循环 (用户手动停止)
+      return { continue: false, action: 'stop_exit' }
+
+    case 'running':
+      // 正常继续
+      return { continue: true, action: 'continue' }
+
+    default:
+      // 异常状态
+      return { continue: false, action: 'stop_exit' }
+  }
+}
+```
+
+### 在 Action 中使用
+
+```markdown
+## Execution
+
+### Step 1: Check Control Signals
+
+\`\`\`javascript
+const control = checkControlSignals(loopId)
+if (!control.continue) {
+  // 输出退出原因
+  console.log(`Loop ${control.action}: status = ${state.status}`)
+
+  // 如果是 pause_exit，保存当前进度
+  if (control.action === 'pause_exit') {
+    updateSkillState(loopId, { current_action: 'paused' })
+  }
+
+  return  // 退出 Action
+}
+\`\`\`
+
+### Step 2: Execute Action Logic
+...
+```
+
+## 状态转换规则
+
+### 1. 初始化 (action-init)
+
+```javascript
+// Skill 初始化后
+{
+  // API 字段更新
+  status: 'created' → 'running',  // 或保持 'running' 如果 API 已设置
+  updated_at: timestamp,
+
+  // Skill 字段初始化
+  skill_state: {
+    current_action: 'init',
+    mode: 'auto',
+    develop: {
+      tasks: [...parsed_tasks],
+      total: N,
+      completed: 0
+    }
+  }
+}
+```
+
+### 2. 开发进行中 (action-develop-with-file)
+
+```javascript
+// 开发任务执行后
+{
+  updated_at: timestamp,
+  current_iteration: state.current_iteration + 1,
+
+  skill_state: {
+    current_action: 'develop',
+    last_action: 'action-develop-with-file',
+    completed_actions: [...state.skill_state.completed_actions, 'action-develop-with-file'],
+    develop: {
+      current_task: 'task-xxx',
+      completed: N+1,
+      last_progress_at: timestamp
+    }
+  }
+}
+```
+
+### 3. 调试进行中 (action-debug-with-file)
+
+```javascript
+// 调试执行后
+{
+  updated_at: timestamp,
+  current_iteration: state.current_iteration + 1,
+
+  skill_state: {
+    current_action: 'debug',
+    last_action: 'action-debug-with-file',
+    debug: {
+      active_bug: '...',
+      hypotheses_count: N,
+      hypotheses: [...new_hypotheses],
+      iteration: N+1,
+      last_analysis_at: timestamp
+    }
+  }
+}
+```
+
+### 4. 验证完成 (action-validate-with-file)
+
+```javascript
+// 验证执行后
+{
+  updated_at: timestamp,
+  current_iteration: state.current_iteration + 1,
+
+  skill_state: {
+    current_action: 'validate',
+    last_action: 'action-validate-with-file',
+    validate: {
+      test_results: [...results],
+      pass_rate: 95.5,
+      coverage: 85.0,
+      passed: true | false,
+      failed_tests: ['test1', 'test2'],
+      last_run_at: timestamp
+    }
+  }
+}
+```
+
+### 5. 完成 (action-complete)
+
+```javascript
+// 循环完成后
+{
+  status: 'running' → 'completed',
+  completed_at: timestamp,
+  updated_at: timestamp,
+
+  skill_state: {
+    current_action: 'complete',
+    last_action: 'action-complete'
+  }
+}
+```
+
+## 状态派生字段
+
+以下字段可从状态计算得出，不需要存储：
+
+```javascript
+// 开发完成度
+const developProgress = state.develop.total_count > 0
+  ? (state.develop.completed_count / state.develop.total_count) * 100
+  : 0
+
+// 是否有待开发任务
+const hasPendingDevelop = state.develop.tasks.some(t => t.status === 'pending')
+
+// 调试是否完成
+const debugCompleted = state.debug.confirmed_hypothesis !== null
+
+// 验证是否通过
+const validationPassed = state.validate.passed && state.validate.test_results.length > 0
+
+// 整体进度
+const overallProgress = (
+  (developProgress * 0.5) +
+  (debugCompleted ? 25 : 0) +
+  (validationPassed ? 25 : 0)
+)
+```
+
+## 文件同步
+
+### 统一位置 (Unified Location)
+
+状态与文件的对应关系：
+
+| 状态字段 | 同步文件 | 同步时机 |
+|----------|----------|----------|
+| 整个 LoopState | `.loop/{loopId}.json` | 每次状态变更 (主文件) |
+| `skill_state.develop` | `.loop/{loopId}.progress/develop.md` | 每次开发操作后 |
+| `skill_state.debug` | `.loop/{loopId}.progress/debug.md` | 每次调试操作后 |
+| `skill_state.validate` | `.loop/{loopId}.progress/validate.md` | 每次验证操作后 |
+| 代码变更日志 | `.loop/{loopId}.progress/changes.log` | 每次文件修改 (NDJSON) |
+| 调试日志 | `.loop/{loopId}.progress/debug.log` | 每次调试日志 (NDJSON) |
+
+### 文件结构示例
+
+```
+.loop/
+├── loop-v2-20260122-abc123.json         # 主状态文件 (API + Skill)
+├── loop-v2-20260122-abc123.tasks.jsonl  # 任务列表 (API 管理)
+└── loop-v2-20260122-abc123.progress/    # Skill 进度文件
+    ├── develop.md                       # 开发进度
+    ├── debug.md                         # 调试理解
+    ├── validate.md                      # 验证报告
+    ├── changes.log                      # 代码变更 (NDJSON)
+    └── debug.log                        # 调试日志 (NDJSON)
+```
+
+## 状态恢复
+
+如果主状态文件 `.loop/{loopId}.json` 损坏，可以从进度文件重建 skill_state:
+
+```javascript
+function rebuildSkillStateFromProgress(loopId) {
+  const progressDir = `.loop/${loopId}.progress`
+
+  // 尝试从进度文件解析状态
+  const skill_state = {
+    develop: parseProgressFile(`${progressDir}/develop.md`),
+    debug: parseProgressFile(`${progressDir}/debug.md`),
+    validate: parseProgressFile(`${progressDir}/validate.md`)
+  }
+
+  return skill_state
+}
+
+// 解析进度 Markdown 文件
+function parseProgressFile(filePath) {
+  const content = Read(filePath)
+  if (!content) return null
+
+  // 从 Markdown 表格和结构中提取数据
+  // ... implementation
+}
+```
+
+### 恢复策略
+
+1. **API 字段**: 无法恢复 - 需要从 API 重新获取或用户手动输入
+2. **skill_state 字段**: 可以从 `.progress/` 目录的 Markdown 文件解析
+3. **任务列表**: 从 `.loop/{loopId}.tasks.jsonl` 恢复
+
+## 状态验证
+
+```javascript
+function validateState(state) {
+  const errors = []
+
+  // 必需字段
+  if (!state.session_id) errors.push('Missing session_id')
+  if (!state.task_description) errors.push('Missing task_description')
+
+  // 状态一致性
+  if (state.initialized && state.status === 'pending') {
+    errors.push('Inconsistent: initialized but status is pending')
+  }
+
+  if (state.status === 'completed' && !state.validate.passed) {
+    errors.push('Inconsistent: completed but validation not passed')
+  }
+
+  // 开发任务一致性
+  const completedTasks = state.develop.tasks.filter(t => t.status === 'completed').length
+  if (completedTasks !== state.develop.completed_count) {
+    errors.push('Inconsistent: completed_count mismatch')
+  }
+
+  return { valid: errors.length === 0, errors }
+}
+```

.claude/skills/ccw-loop/specs/action-catalog.md

@@ -0,0 +1,300 @@
+# Action Catalog
+
+CCW Loop 所有可用动作的目录和说明。
+
+## Available Actions
+
+| Action | Purpose | Preconditions | Effects | CLI Integration |
+|--------|---------|---------------|---------|-----------------|
+| [action-init](../phases/actions/action-init.md) | 初始化会话 | status=pending, initialized=false | status→running, initialized→true, 创建目录和任务列表 | Gemini 任务分解 |
+| [action-menu](../phases/actions/action-menu.md) | 显示操作菜单 | initialized=true, status=running | 返回用户选择的动作 | - |
+| [action-develop-with-file](../phases/actions/action-develop-with-file.md) | 执行开发任务 | initialized=true, pending tasks > 0 | 更新 progress.md, 完成一个任务 | Gemini 代码实现 |
+| [action-debug-with-file](../phases/actions/action-debug-with-file.md) | 假设驱动调试 | initialized=true | 更新 understanding.md, hypotheses.json | Gemini 假设生成和证据分析 |
+| [action-validate-with-file](../phases/actions/action-validate-with-file.md) | 运行测试验证 | initialized=true, develop > 0 or debug confirmed | 更新 validation.md, test-results.json | Gemini 质量分析 |
+| [action-complete](../phases/actions/action-complete.md) | 完成循环 | initialized=true | status→completed, 生成 summary.md | - |
+
+## Action Dependencies Graph
+
+```mermaid
+graph TD
+    START([用户启动 /ccw-loop]) --> INIT[action-init]
+    INIT --> MENU[action-menu]
+
+    MENU --> DEVELOP[action-develop-with-file]
+    MENU --> DEBUG[action-debug-with-file]
+    MENU --> VALIDATE[action-validate-with-file]
+    MENU --> STATUS[action-status]
+    MENU --> COMPLETE[action-complete]
+    MENU --> EXIT([退出])
+
+    DEVELOP --> MENU
+    DEBUG --> MENU
+    VALIDATE --> MENU
+    STATUS --> MENU
+    COMPLETE --> END([结束])
+    EXIT --> END
+
+    style INIT fill:#e1f5fe
+    style MENU fill:#fff3e0
+    style DEVELOP fill:#e8f5e9
+    style DEBUG fill:#fce4ec
+    style VALIDATE fill:#f3e5f5
+    style COMPLETE fill:#c8e6c9
+```
+
+## Action Execution Matrix
+
+### Interactive Mode
+
+| State | Auto-Selected Action | User Options |
+|-------|---------------------|--------------|
+| pending | action-init | - |
+| running, !initialized | action-init | - |
+| running, initialized | action-menu | All actions |
+
+### Auto Mode
+
+| Condition | Selected Action |
+|-----------|----------------|
+| pending_develop_tasks > 0 | action-develop-with-file |
+| last_action=develop, !debug_completed | action-debug-with-file |
+| last_action=debug, !validation_completed | action-validate-with-file |
+| validation_failed | action-develop-with-file (fix) |
+| validation_passed, no pending | action-complete |
+
+## Action Inputs/Outputs
+
+### action-init
+
+**Inputs**:
+- state.task_description
+- User input (optional)
+
+**Outputs**:
+- meta.json
+- state.json (初始化)
+- develop/tasks.json
+- develop/progress.md
+
+**State Changes**:
+```javascript
+{
+  status: 'pending' → 'running',
+  initialized: false → true,
+  develop.tasks: [] → [task1, task2, ...]
+}
+```
+
+### action-develop-with-file
+
+**Inputs**:
+- state.develop.tasks
+- User selection (如有多个待处理任务)
+
+**Outputs**:
+- develop/progress.md (追加)
+- develop/tasks.json (更新)
+- develop/changes.log (追加)
+
+**State Changes**:
+```javascript
+{
+  develop.current_task_id: null → 'task-xxx' → null,
+  develop.completed_count: N → N+1,
+  last_action: X → 'action-develop-with-file'
+}
+```
+
+### action-debug-with-file
+
+**Inputs**:
+- Bug description (用户输入或从测试失败获取)
+- debug.log (如已有)
+
+**Outputs**:
+- debug/understanding.md (追加)
+- debug/hypotheses.json (更新)
+- Code changes (添加日志或修复)
+
+**State Changes**:
+```javascript
+{
+  debug.current_bug: null → 'bug description',
+  debug.hypotheses: [...updated],
+  debug.iteration: N → N+1,
+  debug.confirmed_hypothesis: null → 'H1' (如确认)
+}
+```
+
+### action-validate-with-file
+
+**Inputs**:
+- 测试脚本 (从 package.json)
+- 覆盖率工具 (可选)
+
+**Outputs**:
+- validate/validation.md (追加)
+- validate/test-results.json (更新)
+- validate/coverage.json (更新)
+
+**State Changes**:
+```javascript
+{
+  validate.test_results: [...new results],
+  validate.coverage: null → 85.5,
+  validate.passed: false → true,
+  validate.failed_tests: ['test1', 'test2'] → []
+}
+```
+
+### action-complete
+
+**Inputs**:
+- state (完整状态)
+- User choices (扩展选项)
+
+**Outputs**:
+- summary.md
+- Issues (如选择扩展)
+
+**State Changes**:
+```javascript
+{
+  status: 'running' → 'completed',
+  completed_at: null → timestamp
+}
+```
+
+## Action Sequences
+
+### Typical Happy Path
+
+```
+action-init
+  → action-develop-with-file (task 1)
+  → action-develop-with-file (task 2)
+  → action-develop-with-file (task 3)
+  → action-validate-with-file
+    → PASS
+  → action-complete
+```
+
+### Debug Iteration Path
+
+```
+action-init
+  → action-develop-with-file (task 1)
+  → action-validate-with-file
+    → FAIL
+  → action-debug-with-file (探索)
+  → action-debug-with-file (分析)
+    → Root cause found
+  → action-validate-with-file
+    → PASS
+  → action-complete
+```
+
+### Multi-Iteration Path
+
+```
+action-init
+  → action-develop-with-file (task 1)
+  → action-debug-with-file
+  → action-develop-with-file (task 2)
+  → action-validate-with-file
+    → FAIL
+  → action-debug-with-file
+  → action-validate-with-file
+    → PASS
+  → action-complete
+```
+
+## Error Scenarios
+
+### CLI Tool Failure
+
+```
+action-develop-with-file
+  → Gemini CLI fails
+  → Fallback to manual implementation
+  → Prompt user for code
+  → Continue
+```
+
+### Test Failure
+
+```
+action-validate-with-file
+  → Tests fail
+  → Record failed tests
+  → Suggest action-debug-with-file
+  → User chooses debug or manual fix
+```
+
+### Max Iterations Reached
+
+```
+state.iteration_count >= 10
+  → Warning message
+  → Suggest break or task split
+  → Allow continue or exit
+```
+
+## Action Extensions
+
+### Adding New Actions
+
+To add a new action:
+
+1. Create `phases/actions/action-{name}.md`
+2. Define preconditions, execution, state updates
+3. Add to this catalog
+4. Update orchestrator.md decision logic
+5. Add to action-menu.md options
+
+### Action Template
+
+```markdown
+# Action: {Name}
+
+{Brief description}
+
+## Purpose
+
+{Detailed purpose}
+
+## Preconditions
+
+- [ ] condition1
+- [ ] condition2
+
+## Execution
+
+### Step 1: {Step Name}
+
+\`\`\`javascript
+// code
+\`\`\`
+
+## State Updates
+
+\`\`\`javascript
+return {
+  stateUpdates: {
+    // updates
+  },
+  continue: true,
+  message: "..."
+}
+\`\`\`
+
+## Error Handling
+
+| Error Type | Recovery |
+|------------|----------|
+| ... | ... |
+
+## Next Actions (Hints)
+
+- condition: next_action
+```

.claude/skills/ccw-loop/specs/loop-requirements.md

@@ -0,0 +1,192 @@
+# Loop Requirements Specification
+
+CCW Loop 的核心需求和约束定义。
+
+## Core Requirements
+
+### 1. 无状态循环
+
+**Requirement**: 每次执行从文件读取状态，执行后写回文件，不依赖内存状态。
+
+**Rationale**: 支持随时中断和恢复，状态持久化。
+
+**Validation**:
+- [ ] 每个 action 开始时从文件读取状态
+- [ ] 每个 action 结束时将状态写回文件
+- [ ] 无全局变量或内存状态依赖
+
+### 2. 文件驱动进度
+
+**Requirement**: 所有进度、理解、验证结果都记录在专用 Markdown 文件中。
+
+**Rationale**: 可审计、可回顾、团队可见。
+
+**Validation**:
+- [ ] develop/progress.md 记录开发进度
+- [ ] debug/understanding.md 记录理解演变
+- [ ] validate/validation.md 记录验证结果
+- [ ] 所有文件使用 Markdown 格式，易读
+
+### 3. CLI 工具集成
+
+**Requirement**: 关键决策点使用 Gemini/CLI 进行深度分析。
+
+**Rationale**: 利用 LLM 能力提高质量。
+
+**Validation**:
+- [ ] 任务分解使用 Gemini
+- [ ] 假设生成使用 Gemini
+- [ ] 证据分析使用 Gemini
+- [ ] 质量评估使用 Gemini
+
+### 4. 用户控制循环
+
+**Requirement**: 支持交互式和自动循环两种模式，用户可随时介入。
+
+**Rationale**: 灵活性，适应不同场景。
+
+**Validation**:
+- [ ] 交互模式：每步显示菜单
+- [ ] 自动模式：按预设流程执行
+- [ ] 用户可随时退出
+- [ ] 状态可恢复
+
+### 5. 可恢复性
+
+**Requirement**: 任何时候中断后，可以从上次位置继续。
+
+**Rationale**: 长时间任务支持，意外中断恢复。
+
+**Validation**:
+- [ ] 状态保存在 state.json
+- [ ] 使用 --resume 可继续
+- [ ] 历史记录完整保留
+
+## Quality Standards
+
+### Completeness
+
+| Dimension | Threshold |
+|-----------|-----------|
+| 进度文档完整性 | 每个任务都有记录 |
+| 理解文档演变 | 每次迭代都有更新 |
+| 验证报告详尽 | 包含所有测试结果 |
+
+### Consistency
+
+| Dimension | Threshold |
+|-----------|-----------|
+| 文件格式一致 | 所有 Markdown 文件使用相同模板 |
+| 状态同步一致 | state.json 与文件内容匹配 |
+| 时间戳格式 | 统一使用 ISO8601 格式 |
+
+### Usability
+
+| Dimension | Threshold |
+|-----------|-----------|
+| 菜单易用性 | 选项清晰，描述准确 |
+| 进度可见性 | 随时可查看当前状态 |
+| 错误提示 | 错误消息清晰，提供恢复建议 |
+
+## Constraints
+
+### 1. 文件结构约束
+
+```
+.workflow/.loop/{session-id}/
+├── meta.json           # 只写一次，不再修改
+├── state.json          # 每次 action 后更新
+├── develop/
+│   ├── progress.md     # 只追加，不删除
+│   ├── tasks.json      # 任务状态更新
+│   └── changes.log     # NDJSON 格式，只追加
+├── debug/
+│   ├── understanding.md   # 只追加，记录时间线
+│   ├── hypotheses.json    # 更新假设状态
+│   └── debug.log          # NDJSON 格式
+└── validate/
+    ├── validation.md      # 每次验证追加
+    ├── test-results.json  # 累积测试结果
+    └── coverage.json      # 最新覆盖率
+```
+
+### 2. 命名约束
+
+- Session ID: `LOOP-{slug}-{YYYY-MM-DD}`
+- Task ID: `task-{NNN}` (三位数字)
+- Hypothesis ID: `H{N}` (单字母+数字)
+
+### 3. 状态转换约束
+
+```
+pending → running → completed
+              ↓
+         user_exit
+              ↓
+            failed
+```
+
+Only allow: `pending→running`, `running→completed/user_exit/failed`
+
+### 4. 错误限制约束
+
+- 最大错误次数: 3
+- 超过 3 次错误 → 自动终止
+- 每次错误 → 记录到 state.errors[]
+
+### 5. 迭代限制约束
+
+- 最大迭代次数: 10 (警告)
+- 超过 10 次 → 警告用户，但不强制停止
+- 建议拆分任务或休息
+
+## Integration Requirements
+
+### 1. Dashboard 集成
+
+**Requirement**: 与 CCW Dashboard Loop Monitor 无缝集成。
+
+**Specification**:
+- Dashboard 创建 Loop → 调用此 Skill
+- state.json → Dashboard 实时显示
+- 任务列表双向同步
+- 状态控制按钮映射到 actions
+
+### 2. Issue 系统集成
+
+**Requirement**: 完成后可扩展为 Issue。
+
+**Specification**:
+- 支持维度: test, enhance, refactor, doc
+- 调用 `/issue:new "{summary} - {dimension}"`
+- 自动填充上下文
+
+### 3. CLI 工具集成
+
+**Requirement**: 使用 CCW CLI 工具进行分析和实现。
+
+**Specification**:
+- 任务分解: `--rule planning-breakdown-task-steps`
+- 代码实现: `--rule development-implement-feature`
+- 根因分析: `--rule analysis-diagnose-bug-root-cause`
+- 质量评估: `--rule analysis-review-code-quality`
+
+## Non-Functional Requirements
+
+### Performance
+
+- Session 初始化: < 5s
+- Action 执行: < 30s (不含 CLI 调用)
+- 状态读写: < 1s
+
+### Reliability
+
+- 状态文件损坏恢复: 支持从其他文件重建
+- CLI 工具失败降级: 回退到手动模式
+- 错误重试: 支持一次自动重试
+
+### Maintainability
+
+- 文档化: 所有 action 都有清晰说明
+- 模块化: 每个 action 独立可测
+- 可扩展: 易于添加新 action

.claude/skills/ccw-loop/templates/progress-template.md

@@ -0,0 +1,175 @@
+# Progress Document Template
+
+开发进度文档的标准模板。
+
+## Template Structure
+
+```markdown
+# Development Progress
+
+**Session ID**: {{session_id}}
+**Task**: {{task_description}}
+**Started**: {{started_at}}
+**Estimated Complexity**: {{complexity}}
+
+---
+
+## Task List
+
+{{#each tasks}}
+{{@index}}. [{{#if completed}}x{{else}} {{/if}}] {{description}}
+{{/each}}
+
+## Key Files
+
+{{#each key_files}}
+- `{{this}}`
+{{/each}}
+
+---
+
+## Progress Timeline
+
+{{#each iterations}}
+### Iteration {{@index}} - {{task_name}} ({{timestamp}})
+
+#### Task Details
+
+- **ID**: {{task_id}}
+- **Tool**: {{tool}}
+- **Mode**: {{mode}}
+
+#### Implementation Summary
+
+{{summary}}
+
+#### Files Changed
+
+{{#each files_changed}}
+- `{{this}}`
+{{/each}}
+
+#### Status: {{status}}
+
+---
+{{/each}}
+
+## Current Statistics
+
+| Metric | Value |
+|--------|-------|
+| Total Tasks | {{total_tasks}} |
+| Completed | {{completed_tasks}} |
+| In Progress | {{in_progress_tasks}} |
+| Pending | {{pending_tasks}} |
+| Progress | {{progress_percentage}}% |
+
+---
+
+## Next Steps
+
+{{#each next_steps}}
+- [ ] {{this}}
+{{/each}}
+```
+
+## Template Variables
+
+| Variable | Type | Source | Description |
+|----------|------|--------|-------------|
+| `session_id` | string | state.session_id | 会话 ID |
+| `task_description` | string | state.task_description | 任务描述 |
+| `started_at` | string | state.created_at | 开始时间 |
+| `complexity` | string | state.context.estimated_complexity | 预估复杂度 |
+| `tasks` | array | state.develop.tasks | 任务列表 |
+| `key_files` | array | state.context.key_files | 关键文件 |
+| `iterations` | array | 从文件解析 | 迭代历史 |
+| `total_tasks` | number | state.develop.total_count | 总任务数 |
+| `completed_tasks` | number | state.develop.completed_count | 已完成数 |
+
+## Usage Example
+
+```javascript
+const progressTemplate = Read('.claude/skills/ccw-loop/templates/progress-template.md')
+
+function renderProgress(state) {
+  let content = progressTemplate
+
+  // 替换简单变量
+  content = content.replace('{{session_id}}', state.session_id)
+  content = content.replace('{{task_description}}', state.task_description)
+  content = content.replace('{{started_at}}', state.created_at)
+  content = content.replace('{{complexity}}', state.context?.estimated_complexity || 'unknown')
+
+  // 替换任务列表
+  const taskList = state.develop.tasks.map((t, i) => {
+    const checkbox = t.status === 'completed' ? 'x' : ' '
+    return `${i + 1}. [${checkbox}] ${t.description}`
+  }).join('\n')
+  content = content.replace('{{#each tasks}}...{{/each}}', taskList)
+
+  // 替换统计
+  content = content.replace('{{total_tasks}}', state.develop.total_count)
+  content = content.replace('{{completed_tasks}}', state.develop.completed_count)
+  // ...
+
+  return content
+}
+```
+
+## Section Templates
+
+### Task Entry
+
+```markdown
+### Iteration {{N}} - {{task_name}} ({{timestamp}})
+
+#### Task Details
+
+- **ID**: {{task_id}}
+- **Tool**: {{tool}}
+- **Mode**: {{mode}}
+
+#### Implementation Summary
+
+{{summary}}
+
+#### Files Changed
+
+{{#each files}}
+- `{{this}}`
+{{/each}}
+
+#### Status: COMPLETED
+
+---
+```
+
+### Statistics Table
+
+```markdown
+## Current Statistics
+
+| Metric | Value |
+|--------|-------|
+| Total Tasks | {{total}} |
+| Completed | {{completed}} |
+| In Progress | {{in_progress}} |
+| Pending | {{pending}} |
+| Progress | {{percentage}}% |
+```
+
+### Next Steps
+
+```markdown
+## Next Steps
+
+{{#if all_completed}}
+- [ ] Run validation tests
+- [ ] Code review
+- [ ] Update documentation
+{{else}}
+- [ ] Complete remaining {{pending}} tasks
+- [ ] Review completed work
+{{/if}}
+```

.claude/skills/ccw-loop/templates/understanding-template.md

@@ -0,0 +1,303 @@
+# Understanding Document Template
+
+调试理解演变文档的标准模板。
+
+## Template Structure
+
+```markdown
+# Understanding Document
+
+**Session ID**: {{session_id}}
+**Bug Description**: {{bug_description}}
+**Started**: {{started_at}}
+
+---
+
+## Exploration Timeline
+
+{{#each iterations}}
+### Iteration {{number}} - {{title}} ({{timestamp}})
+
+{{#if is_exploration}}
+#### Current Understanding
+
+Based on bug description and initial code search:
+
+- Error pattern: {{error_pattern}}
+- Affected areas: {{affected_areas}}
+- Initial hypothesis: {{initial_thoughts}}
+
+#### Evidence from Code Search
+
+{{#each search_results}}
+**Keyword: "{{keyword}}"**
+- Found in: {{files}}
+- Key findings: {{insights}}
+{{/each}}
+{{/if}}
+
+{{#if has_hypotheses}}
+#### Hypotheses Generated (Gemini-Assisted)
+
+{{#each hypotheses}}
+**{{id}}** (Likelihood: {{likelihood}}): {{description}}
+- Logging at: {{logging_point}}
+- Testing: {{testable_condition}}
+- Evidence to confirm: {{confirm_criteria}}
+- Evidence to reject: {{reject_criteria}}
+{{/each}}
+
+**Gemini Insights**: {{gemini_insights}}
+{{/if}}
+
+{{#if is_analysis}}
+#### Log Analysis Results
+
+{{#each results}}
+**{{id}}**: {{verdict}}
+- Evidence: {{evidence}}
+- Reasoning: {{reason}}
+{{/each}}
+
+#### Corrected Understanding
+
+Previous misunderstandings identified and corrected:
+
+{{#each corrections}}
+- ~~{{wrong}}~~ → {{corrected}}
+  - Why wrong: {{reason}}
+  - Evidence: {{evidence}}
+{{/each}}
+
+#### New Insights
+
+{{#each insights}}
+- {{this}}
+{{/each}}
+
+#### Gemini Analysis
+
+{{gemini_analysis}}
+{{/if}}
+
+{{#if root_cause_found}}
+#### Root Cause Identified
+
+**{{hypothesis_id}}**: {{description}}
+
+Evidence supporting this conclusion:
+{{supporting_evidence}}
+{{else}}
+#### Next Steps
+
+{{next_steps}}
+{{/if}}
+
+---
+{{/each}}
+
+## Current Consolidated Understanding
+
+### What We Know
+
+{{#each valid_understandings}}
+- {{this}}
+{{/each}}
+
+### What Was Disproven
+
+{{#each disproven}}
+- ~~{{assumption}}~~ (Evidence: {{evidence}})
+{{/each}}
+
+### Current Investigation Focus
+
+{{current_focus}}
+
+### Remaining Questions
+
+{{#each questions}}
+- {{this}}
+{{/each}}
+```
+
+## Template Variables
+
+| Variable | Type | Source | Description |
+|----------|------|--------|-------------|
+| `session_id` | string | state.session_id | 会话 ID |
+| `bug_description` | string | state.debug.current_bug | Bug 描述 |
+| `iterations` | array | 从文件解析 | 迭代历史 |
+| `hypotheses` | array | state.debug.hypotheses | 假设列表 |
+| `valid_understandings` | array | 从 Gemini 分析 | 有效理解 |
+| `disproven` | array | 从假设状态 | 被否定的假设 |
+
+## Section Templates
+
+### Exploration Section
+
+```markdown
+### Iteration {{N}} - Initial Exploration ({{timestamp}})
+
+#### Current Understanding
+
+Based on bug description and initial code search:
+
+- Error pattern: {{pattern}}
+- Affected areas: {{areas}}
+- Initial hypothesis: {{thoughts}}
+
+#### Evidence from Code Search
+
+{{#each search_results}}
+**Keyword: "{{keyword}}"**
+- Found in: {{files}}
+- Key findings: {{insights}}
+{{/each}}
+
+#### Next Steps
+
+- Generate testable hypotheses
+- Add instrumentation
+- Await reproduction
+```
+
+### Hypothesis Section
+
+```markdown
+#### Hypotheses Generated (Gemini-Assisted)
+
+| ID | Description | Likelihood | Status |
+|----|-------------|------------|--------|
+{{#each hypotheses}}
+| {{id}} | {{description}} | {{likelihood}} | {{status}} |
+{{/each}}
+
+**Details:**
+
+{{#each hypotheses}}
+**{{id}}**: {{description}}
+- Logging at: `{{logging_point}}`
+- Testing: {{testable_condition}}
+- Confirm: {{evidence_criteria.confirm}}
+- Reject: {{evidence_criteria.reject}}
+{{/each}}
+```
+
+### Analysis Section
+
+```markdown
+### Iteration {{N}} - Evidence Analysis ({{timestamp}})
+
+#### Log Analysis Results
+
+{{#each results}}
+**{{id}}**: **{{verdict}}**
+- Evidence: \`{{evidence}}\`
+- Reasoning: {{reason}}
+{{/each}}
+
+#### Corrected Understanding
+
+| Previous Assumption | Corrected To | Reason |
+|---------------------|--------------|--------|
+{{#each corrections}}
+| ~~{{wrong}}~~ | {{corrected}} | {{reason}} |
+{{/each}}
+
+#### Gemini Analysis
+
+{{gemini_analysis}}
+```
+
+### Consolidated Understanding Section
+
+```markdown
+## Current Consolidated Understanding
+
+### What We Know
+
+{{#each valid}}
+- {{this}}
+{{/each}}
+
+### What Was Disproven
+
+{{#each disproven}}
+- ~~{{this.assumption}}~~ (Evidence: {{this.evidence}})
+{{/each}}
+
+### Current Investigation Focus
+
+{{focus}}
+
+### Remaining Questions
+
+{{#each questions}}
+- {{this}}
+{{/each}}
+```
+
+### Resolution Section
+
+```markdown
+### Resolution ({{timestamp}})
+
+#### Fix Applied
+
+- Modified files: {{files}}
+- Fix description: {{description}}
+- Root cause addressed: {{root_cause}}
+
+#### Verification Results
+
+{{verification}}
+
+#### Lessons Learned
+
+{{#each lessons}}
+{{@index}}. {{this}}
+{{/each}}
+
+#### Key Insights for Future
+
+{{#each insights}}
+- {{this}}
+{{/each}}
+```
+
+## Consolidation Rules
+
+更新 "Current Consolidated Understanding" 时遵循以下规则:
+
+1. **简化被否定项**: 移到 "What Was Disproven"，只保留单行摘要
+2. **保留有效见解**: 将确认的发现提升到 "What We Know"
+3. **避免重复**: 不在合并部分重复时间线细节
+4. **关注当前状态**: 描述现在知道什么，而不是过程
+5. **保留关键纠正**: 保留重要的 wrong→right 转换供学习
+
+## Anti-Patterns
+
+**错误示例 (冗余)**:
+```markdown
+## Current Consolidated Understanding
+
+In iteration 1 we thought X, but in iteration 2 we found Y, then in iteration 3...
+Also we checked A and found B, and then we checked C...
+```
+
+**正确示例 (精简)**:
+```markdown
+## Current Consolidated Understanding
+
+### What We Know
+- Error occurs during runtime update, not initialization
+- Config value is None (not missing key)
+
+### What Was Disproven
+- ~~Initialization error~~ (Timing evidence)
+- ~~Missing key hypothesis~~ (Key exists)
+
+### Current Investigation Focus
+Why is config value None during update?
+```

.claude/skills/ccw-loop/templates/validation-template.md

@@ -0,0 +1,258 @@
+# Validation Report Template
+
+验证报告的标准模板。
+
+## Template Structure
+
+```markdown
+# Validation Report
+
+**Session ID**: {{session_id}}
+**Task**: {{task_description}}
+**Validated**: {{timestamp}}
+
+---
+
+## Iteration {{iteration}} - Validation Run
+
+### Test Execution Summary
+
+| Metric | Value |
+|--------|-------|
+| Total Tests | {{total_tests}} |
+| Passed | {{passed_tests}} |
+| Failed | {{failed_tests}} |
+| Skipped | {{skipped_tests}} |
+| Duration | {{duration}}ms |
+| **Pass Rate** | **{{pass_rate}}%** |
+
+### Coverage Report
+
+{{#if has_coverage}}
+| File | Statements | Branches | Functions | Lines |
+|------|------------|----------|-----------|-------|
+{{#each coverage_files}}
+| {{path}} | {{statements}}% | {{branches}}% | {{functions}}% | {{lines}}% |
+{{/each}}
+
+**Overall Coverage**: {{overall_coverage}}%
+{{else}}
+_No coverage data available_
+{{/if}}
+
+### Failed Tests
+
+{{#if has_failures}}
+{{#each failures}}
+#### {{test_name}}
+
+- **Suite**: {{suite}}
+- **Error**: {{error_message}}
+- **Stack**:
+\`\`\`
+{{stack_trace}}
+\`\`\`
+{{/each}}
+{{else}}
+_All tests passed_
+{{/if}}
+
+### Gemini Quality Analysis
+
+{{gemini_analysis}}
+
+### Recommendations
+
+{{#each recommendations}}
+- {{this}}
+{{/each}}
+
+---
+
+## Validation Decision
+
+**Result**: {{#if passed}}✅ PASS{{else}}❌ FAIL{{/if}}
+
+**Rationale**: {{rationale}}
+
+{{#if not_passed}}
+### Next Actions
+
+1. Review failed tests
+2. Debug failures using action-debug-with-file
+3. Fix issues and re-run validation
+{{else}}
+### Next Actions
+
+1. Consider code review
+2. Prepare for deployment
+3. Update documentation
+{{/if}}
+```
+
+## Template Variables
+
+| Variable | Type | Source | Description |
+|----------|------|--------|-------------|
+| `session_id` | string | state.session_id | 会话 ID |
+| `task_description` | string | state.task_description | 任务描述 |
+| `timestamp` | string | 当前时间 | 验证时间 |
+| `iteration` | number | 从文件计算 | 验证迭代次数 |
+| `total_tests` | number | 测试输出 | 总测试数 |
+| `passed_tests` | number | 测试输出 | 通过数 |
+| `failed_tests` | number | 测试输出 | 失败数 |
+| `pass_rate` | number | 计算得出 | 通过率 |
+| `coverage_files` | array | 覆盖率报告 | 文件覆盖率 |
+| `failures` | array | 测试输出 | 失败测试详情 |
+| `gemini_analysis` | string | Gemini CLI | 质量分析 |
+| `recommendations` | array | Gemini CLI | 建议列表 |
+
+## Section Templates
+
+### Test Summary
+
+```markdown
+### Test Execution Summary
+
+| Metric | Value |
+|--------|-------|
+| Total Tests | {{total}} |
+| Passed | {{passed}} |
+| Failed | {{failed}} |
+| Skipped | {{skipped}} |
+| Duration | {{duration}}ms |
+| **Pass Rate** | **{{rate}}%** |
+```
+
+### Coverage Table
+
+```markdown
+### Coverage Report
+
+| File | Statements | Branches | Functions | Lines |
+|------|------------|----------|-----------|-------|
+{{#each files}}
+| `{{path}}` | {{statements}}% | {{branches}}% | {{functions}}% | {{lines}}% |
+{{/each}}
+
+**Overall Coverage**: {{overall}}%
+
+**Coverage Thresholds**:
+- ✅ Good: ≥ 80%
+- ⚠️ Warning: 60-79%
+- ❌ Poor: < 60%
+```
+
+### Failed Test Details
+
+```markdown
+### Failed Tests
+
+{{#each failures}}
+#### ❌ {{test_name}}
+
+| Field | Value |
+|-------|-------|
+| Suite | {{suite}} |
+| Error | {{error_message}} |
+| Duration | {{duration}}ms |
+
+**Stack Trace**:
+\`\`\`
+{{stack_trace}}
+\`\`\`
+
+**Possible Causes**:
+{{#each possible_causes}}
+- {{this}}
+{{/each}}
+
+---
+{{/each}}
+```
+
+### Quality Analysis
+
+```markdown
+### Gemini Quality Analysis
+
+#### Code Quality Assessment
+
+| Dimension | Score | Status |
+|-----------|-------|--------|
+| Correctness | {{correctness}}/10 | {{correctness_status}} |
+| Completeness | {{completeness}}/10 | {{completeness_status}} |
+| Reliability | {{reliability}}/10 | {{reliability_status}} |
+| Maintainability | {{maintainability}}/10 | {{maintainability_status}} |
+
+#### Key Findings
+
+{{#each findings}}
+- **{{severity}}**: {{description}}
+{{/each}}
+
+#### Recommendations
+
+{{#each recommendations}}
+{{@index}}. {{this}}
+{{/each}}
+```
+
+### Decision Section
+
+```markdown
+## Validation Decision
+
+**Result**: {{#if passed}}✅ PASS{{else}}❌ FAIL{{/if}}
+
+**Rationale**:
+{{rationale}}
+
+**Confidence Level**: {{confidence}}
+
+### Decision Matrix
+
+| Criteria | Status | Weight | Score |
+|----------|--------|--------|-------|
+| All tests pass | {{tests_pass}} | 40% | {{tests_score}} |
+| Coverage ≥ 80% | {{coverage_pass}} | 30% | {{coverage_score}} |
+| No critical issues | {{no_critical}} | 20% | {{critical_score}} |
+| Quality analysis pass | {{quality_pass}} | 10% | {{quality_score}} |
+| **Total** | | 100% | **{{total_score}}** |
+
+**Threshold**: 70% to pass
+
+### Next Actions
+
+{{#if passed}}
+1. ✅ Code review (recommended)
+2. ✅ Update documentation
+3. ✅ Prepare for deployment
+{{else}}
+1. ❌ Review failed tests
+2. ❌ Debug failures
+3. ❌ Fix issues and re-run
+{{/if}}
+```
+
+## Historical Comparison
+
+```markdown
+## Validation History
+
+| Iteration | Date | Pass Rate | Coverage | Status |
+|-----------|------|-----------|----------|--------|
+{{#each history}}
+| {{iteration}} | {{date}} | {{pass_rate}}% | {{coverage}}% | {{status}} |
+{{/each}}
+
+### Trend Analysis
+
+{{#if improving}}
+📈 **Improving**: Pass rate increased from {{previous_rate}}% to {{current_rate}}%
+{{else if declining}}
+📉 **Declining**: Pass rate decreased from {{previous_rate}}% to {{current_rate}}%
+{{else}}
+➡️ **Stable**: Pass rate remains at {{current_rate}}%
+{{/if}}
+```

.claude/skills/ccw/SKILL.md

@@ -8,6 +8,44 @@ allowed-tools: Task(*), SlashCommand(*), AskUserQuestion(*), Read(*), Bash(*), G
 
 无状态工作流协调器，根据任务意图自动选择最优工作流。
 
+## Workflow System Overview
+
+CCW 提供两个工作流系统：**Main Workflow** 和 **Issue Workflow**，协同覆盖完整的软件开发生命周期。
+
+```
+┌─────────────────────────────────────────────────────────────────────────────┐
+│                              Main Workflow                                  │
+│                                                                             │
+│  ┌─────────────┐   ┌─────────────┐   ┌─────────────┐   ┌─────────────┐     │
+│  │   Level 1   │ → │   Level 2   │ → │   Level 3   │ → │   Level 4   │     │
+│  │   Rapid     │   │ Lightweight │   │  Standard   │   │ Brainstorm  │     │
+│  │             │   │             │   │             │   │             │     │
+│  │ lite-lite-  │   │ lite-plan   │   │    plan     │   │ brainstorm  │     │
+│  │    lite     │   │ lite-fix    │   │  tdd-plan   │   │  :auto-     │     │
+│  │             │   │ multi-cli-  │   │ test-fix-   │   │  parallel   │     │
+│  │             │   │    plan     │   │    gen      │   │     ↓       │     │
+│  │             │   │             │   │             │   │   plan      │     │
+│  └─────────────┘   └─────────────┘   └─────────────┘   └─────────────┘     │
+│                                                                             │
+│  Complexity: ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━▶  │
+│              Low                                                    High    │
+└─────────────────────────────────────────────────────────────────────────────┘
+                                    │
+                                    │ After development
+                                    ▼
+┌─────────────────────────────────────────────────────────────────────────────┐
+│                              Issue Workflow                                 │
+│                                                                             │
+│     ┌──────────────┐         ┌──────────────┐         ┌──────────────┐     │
+│     │  Accumulate  │    →    │    Plan      │    →    │   Execute    │     │
+│     │  Discover &  │         │    Batch     │         │   Parallel   │     │
+│     │   Collect    │         │   Planning   │         │  Execution   │     │
+│     └──────────────┘         └──────────────┘         └──────────────┘     │
+│                                                                             │
+│     Supplementary role: Maintain main branch stability, worktree isolation  │
+└─────────────────────────────────────────────────────────────────────────────┘
+```
+
 ## Architecture
 
 ```
@@ -17,7 +55,7 @@ allowed-tools: Task(*), SlashCommand(*), AskUserQuestion(*), Read(*), Bash(*), G
 │  Phase 1    │ Input Analysis (rule-based, fast path)            │
 │  Phase 1.5  │ CLI Classification (semantic, smart path)         │
 │  Phase 1.75 │ Requirement Clarification (clarity < 2)           │
-│  Phase 2    │ Chain Selection (intent → workflow)               │
+│  Phase 2    │ Level Selection (intent → level → workflow)       │
 │  Phase 2.5  │ CLI Action Planning (high complexity)             │
 │  Phase 3    │ User Confirmation (optional)                      │
 │  Phase 4    │ TODO Tracking Setup                               │
@@ -25,23 +63,79 @@ allowed-tools: Task(*), SlashCommand(*), AskUserQuestion(*), Read(*), Bash(*), G
 └─────────────────────────────────────────────────────────────────┘
 ```
 
+## Level Quick Reference
+
+| Level | Name | Workflows | Artifacts | Execution |
+|-------|------|-----------|-----------|-----------|
+| **1** | Rapid | `lite-lite-lite` | None | Direct execute |
+| **2** | Lightweight | `lite-plan`, `lite-fix`, `multi-cli-plan` | Memory/Lightweight files | → `lite-execute` |
+| **3** | Standard | `plan`, `tdd-plan`, `test-fix-gen` | Session persistence | → `execute` / `test-cycle-execute` |
+| **4** | Brainstorm | `brainstorm:auto-parallel` → `plan` | Multi-role analysis + Session | → `execute` |
+| **-** | Issue | `discover` → `plan` → `queue` → `execute` | Issue records | Worktree isolation (optional) |
+
+## Workflow Selection Decision Tree
+
+```
+Start
+  │
+  ├─ Is it post-development maintenance?
+  │     ├─ Yes → Issue Workflow
+  │     └─ No ↓
+  │
+  ├─ Are requirements clear?
+  │     ├─ Uncertain → Level 4 (brainstorm:auto-parallel)
+  │     └─ Clear ↓
+  │
+  ├─ Need persistent Session?
+  │     ├─ Yes → Level 3 (plan / tdd-plan / test-fix-gen)
+  │     └─ No ↓
+  │
+  ├─ Need multi-perspective / solution comparison?
+  │     ├─ Yes → Level 2 (multi-cli-plan)
+  │     └─ No ↓
+  │
+  ├─ Is it a bug fix?
+  │     ├─ Yes → Level 2 (lite-fix)
+  │     └─ No ↓
+  │
+  ├─ Need planning?
+  │     ├─ Yes → Level 2 (lite-plan)
+  │     └─ No → Level 1 (lite-lite-lite)
+```
+
 ## Intent Classification
 
-### Priority Order
+### Priority Order (with Level Mapping)
 
-| Priority | Intent | Patterns | Flow |
-|----------|--------|----------|------|
-| 1 | bugfix/hotfix | `urgent,production,critical` + bug | `bugfix.hotfix` |
-| 1 | bugfix | `fix,bug,error,crash,fail` | `bugfix.standard` |
-| 2 | issue batch | `issues,batch` + `fix,resolve` | `issue` |
-| 3 | exploration | `不确定,explore,研究,what if` | `full` |
-| 3 | multi-perspective | `多视角,权衡,比较方案,cross-verify` | `multi-cli-plan` |
-| 4 | quick-task | `快速,简单,small,quick` + feature | `lite-lite-lite` |
-| 5 | ui design | `ui,design,component,style` | `ui` |
-| 6 | tdd | `tdd,test-driven,先写测试` | `tdd` |
-| 7 | review | `review,审查,code review` | `review-fix` |
-| 8 | documentation | `文档,docs,readme` | `docs` |
-| 99 | feature | complexity-based | `rapid`/`coupled` |
+| Priority | Intent | Patterns | Level | Flow |
+|----------|--------|----------|-------|------|
+| 1 | bugfix/hotfix | `urgent,production,critical` + bug | L2 | `bugfix.hotfix` |
+| 1 | bugfix | `fix,bug,error,crash,fail` | L2 | `bugfix.standard` |
+| 2 | issue batch | `issues,batch` + `fix,resolve` | Issue | `issue` |
+| 3 | exploration | `不确定,explore,研究,what if` | L4 | `full` |
+| 3 | multi-perspective | `多视角,权衡,比较方案,cross-verify` | L2 | `multi-cli-plan` |
+| 4 | quick-task | `快速,简单,small,quick` + feature | L1 | `lite-lite-lite` |
+| 5 | ui design | `ui,design,component,style` | L3/L4 | `ui` |
+| 6 | tdd | `tdd,test-driven,先写测试` | L3 | `tdd` |
+| 7 | test-fix | `测试失败,test fail,fix test` | L3 | `test-fix-gen` |
+| 8 | review | `review,审查,code review` | L3 | `review-fix` |
+| 9 | documentation | `文档,docs,readme` | L2 | `docs` |
+| 99 | feature | complexity-based | L2/L3 | `rapid`/`coupled` |
+
+### Quick Selection Guide
+
+| Scenario | Recommended Workflow | Level |
+|----------|---------------------|-------|
+| Quick fixes, config adjustments | `lite-lite-lite` | 1 |
+| Clear single-module features | `lite-plan → lite-execute` | 2 |
+| Bug diagnosis and fix | `lite-fix` | 2 |
+| Production emergencies | `lite-fix --hotfix` | 2 |
+| Technology selection, solution comparison | `multi-cli-plan → lite-execute` | 2 |
+| Multi-module changes, refactoring | `plan → verify → execute` | 3 |
+| Test-driven development | `tdd-plan → execute → tdd-verify` | 3 |
+| Test failure fixes | `test-fix-gen → test-cycle-execute` | 3 |
+| New features, architecture design | `brainstorm:auto-parallel → plan → execute` | 4 |
+| Post-development issue fixes | Issue Workflow | - |
 
 ### Complexity Assessment
 
@@ -214,24 +308,100 @@ CLI 可返回建议：`use_default` | `modify` (调整步骤) | `upgrade` (升
 
 ## Workflow Flow Details
 
-### Issue Workflow (两阶段生命周期)
+### Issue Workflow (Main Workflow 补充机制)
 
-Issue 工作流设计为两阶段生命周期，支持在项目迭代过程中积累问题并集中解决。
+Issue Workflow 是 Main Workflow 的**补充机制**，专注于开发后的持续维护。
 
-**Phase 1: Accumulation (积累阶段)**
-- 触发：任务完成后的 review、代码审查发现、测试失败
-- 活动：需求扩展、bug 分析、测试覆盖、安全审查
-- 命令：`/issue:discover`, `/issue:discover-by-prompt`, `/issue:new`
+#### 设计理念
 
-**Phase 2: Batch Resolution (批量解决阶段)**
-- 触发：积累足够 issue 后的集中处理
-- 流程：plan → queue → execute
-- 命令：`/issue:plan --all-pending` → `/issue:queue` → `/issue:execute`
+| 方面 | Main Workflow | Issue Workflow |
+|------|---------------|----------------|
+| **用途** | 主要开发周期 | 开发后维护 |
+| **时机** | 功能开发阶段 | 主工作流完成后 |
+| **范围** | 完整功能实现 | 针对性修复/增强 |
+| **并行性** | 依赖分析 → Agent 并行 | Worktree 隔离 (可选) |
+| **分支模型** | 当前分支工作 | 可使用隔离的 worktree |
+
+#### 为什么 Main Workflow 不自动使用 Worktree？
+
+**依赖分析已解决并行性问题**：
+1. 规划阶段 (`/workflow:plan`) 执行依赖分析
+2. 自动识别任务依赖和关键路径
+3. 划分为**并行组**（独立任务）和**串行链**（依赖任务）
+4. Agent 并行执行独立任务，无需文件系统隔离
+
+#### 两阶段生命周期
 
 ```
-任务完成 → discover → 积累 issue → ... → plan all → queue → parallel execute
-    ↑                      ↓
-    └────── 迭代循环 ───────┘
+┌─────────────────────────────────────────────────────────────────────┐
+│                    Phase 1: Accumulation (积累阶段)                  │
+│                                                                     │
+│   Triggers: 任务完成后的 review、代码审查发现、测试失败              │
+│                                                                     │
+│   ┌────────────┐     ┌────────────┐     ┌────────────┐             │
+│   │ discover   │     │ discover-  │     │    new     │             │
+│   │ Auto-find  │     │ by-prompt  │     │  Manual    │             │
+│   └────────────┘     └────────────┘     └────────────┘             │
+│                                                                     │
+│   持续积累 issues 到待处理队列                                       │
+└─────────────────────────────────────────────────────────────────────┘
+                               │
+                               │ 积累足够后
+                               ▼
+┌─────────────────────────────────────────────────────────────────────┐
+│                  Phase 2: Batch Resolution (批量解决阶段)            │
+│                                                                     │
+│   ┌────────────┐     ┌────────────┐     ┌────────────┐             │
+│   │   plan     │ ──→ │   queue    │ ──→ │  execute   │             │
+│   │ --all-     │     │ Optimize   │     │  Parallel  │             │
+│   │  pending   │     │  order     │     │ execution  │             │
+│   └────────────┘     └────────────┘     └────────────┘             │
+│                                                                     │
+│   支持 worktree 隔离，保持主分支稳定                                 │
+└─────────────────────────────────────────────────────────────────────┘
+```
+
+#### 与 Main Workflow 的协作
+
+```
+开发迭代循环
+┌─────────────────────────────────────────────────────────────────────┐
+│                                                                     │
+│   ┌─────────┐                              ┌─────────┐             │
+│   │ Feature │ ──→ Main Workflow ──→ Done ──→│ Review  │             │
+│   │ Request │     (Level 1-4)              └────┬────┘             │
+│   └─────────┘                                   │                  │
+│        ▲                                        │ 发现 Issues       │
+│        │                                        ▼                  │
+│        │                                  ┌─────────┐              │
+│   继续 │                                  │  Issue  │              │
+│   新功能│                                  │ Workflow│              │
+│        │                                  └────┬────┘              │
+│        │         ┌──────────────────────────────┘                   │
+│        │         │ 修复完成                                          │
+│        │         ▼                                                 │
+│   ┌────┴────┐◀──────                                               │
+│   │  Main   │    Merge                                             │
+│   │ Branch  │    back                                              │
+│   └─────────┘                                                      │
+│                                                                     │
+└─────────────────────────────────────────────────────────────────────┘
+```
+
+#### 命令列表
+
+**积累阶段：**
+```bash
+/issue:discover            # 多视角自动发现
+/issue:discover-by-prompt  # 基于提示发现
+/issue:new                 # 手动创建
+```
+
+**批量解决阶段：**
+```bash
+/issue:plan --all-pending  # 批量规划所有待处理
+/issue:queue               # 生成优化执行队列
+/issue:execute             # 并行执行
 ```
 
 ### lite-lite-lite vs multi-cli-plan

.claude/skills/ccw/command.json

@@ -73,10 +73,37 @@
   },
 
   "flows": {
+    "_level_guide": {
+      "L1": "Rapid - No artifacts, direct execution",
+      "L2": "Lightweight - Memory/lightweight files, → lite-execute",
+      "L3": "Standard - Session persistence, → execute/test-cycle-execute",
+      "L4": "Brainstorm - Multi-role analysis + Session, → execute"
+    },
+    "lite-lite-lite": {
+      "name": "Ultra-Rapid Execution",
+      "level": "L1",
+      "description": "零文件 + 自动CLI选择 + 语义描述 + 直接执行",
+      "complexity": ["low"],
+      "artifacts": "none",
+      "steps": [
+        { "phase": "clarify", "description": "需求澄清 (AskUser if needed)" },
+        { "phase": "auto-select", "description": "任务分析 → 自动选择CLI组合" },
+        { "phase": "multi-cli", "description": "并行多CLI分析" },
+        { "phase": "decision", "description": "展示结果 → AskUser决策" },
+        { "phase": "execute", "description": "直接执行 (无中间文件)" }
+      ],
+      "cli_hints": {
+        "analysis": { "tool": "auto", "mode": "analysis", "parallel": true },
+        "execution": { "tool": "auto", "mode": "write" }
+      },
+      "estimated_time": "10-30 min"
+    },
     "rapid": {
       "name": "Rapid Iteration",
-      "description": "多模型协作分析 + 直接执行",
+      "level": "L2",
+      "description": "内存规划 + 直接执行",
       "complexity": ["low", "medium"],
+      "artifacts": "memory://plan",
       "steps": [
         { "command": "/workflow:lite-plan", "optional": false, "auto_continue": true },
         { "command": "/workflow:lite-execute", "optional": false }
@@ -87,107 +114,12 @@
       },
       "estimated_time": "15-45 min"
     },
-    "full": {
-      "name": "Full Exploration",
-      "description": "头脑风暴 + 规划 + 执行",
-      "complexity": ["medium", "high"],
-      "steps": [
-        { "command": "/workflow:brainstorm:auto-parallel", "optional": false, "confirm_before": true },
-        { "command": "/workflow:plan", "optional": false },
-        { "command": "/workflow:action-plan-verify", "optional": true, "auto_continue": true },
-        { "command": "/workflow:execute", "optional": false }
-      ],
-      "cli_hints": {
-        "role_analysis": { "tool": "gemini", "mode": "analysis", "trigger": "always", "parallel": true },
-        "execution": { "tool": "codex", "mode": "write", "trigger": "task_count >= 3" }
-      },
-      "estimated_time": "1-3 hours"
-    },
-    "coupled": {
-      "name": "Coupled Planning",
-      "description": "完整规划 + 验证 + 执行",
-      "complexity": ["high"],
-      "steps": [
-        { "command": "/workflow:plan", "optional": false },
-        { "command": "/workflow:action-plan-verify", "optional": false, "auto_continue": true },
-        { "command": "/workflow:execute", "optional": false },
-        { "command": "/workflow:review", "optional": true }
-      ],
-      "cli_hints": {
-        "pre_analysis": { "tool": "gemini", "mode": "analysis", "trigger": "always" },
-        "execution": { "tool": "codex", "mode": "write", "trigger": "always" }
-      },
-      "estimated_time": "2-4 hours"
-    },
-    "bugfix": {
-      "name": "Bug Fix",
-      "description": "智能诊断 + 修复",
-      "complexity": ["low", "medium"],
-      "variants": {
-        "standard": [{ "command": "/workflow:lite-fix", "optional": false }],
-        "hotfix": [{ "command": "/workflow:lite-fix --hotfix", "optional": false }]
-      },
-      "cli_hints": {
-        "diagnosis": { "tool": "gemini", "mode": "analysis", "trigger": "always" },
-        "fix": { "tool": "codex", "mode": "write", "trigger": "severity >= medium" }
-      },
-      "estimated_time": "10-30 min"
-    },
-    "issue": {
-      "name": "Issue Lifecycle",
-      "description": "发现积累 → 批量规划 → 队列优化 → 并行执行",
-      "complexity": ["medium", "high"],
-      "phases": {
-        "accumulation": {
-          "description": "项目迭代中持续发现和积累issue",
-          "commands": ["/issue:discover", "/issue:new"],
-          "trigger": "post-task, code-review, test-failure"
-        },
-        "resolution": {
-          "description": "集中规划和执行积累的issue",
-          "steps": [
-            { "command": "/issue:plan --all-pending", "optional": false },
-            { "command": "/issue:queue", "optional": false },
-            { "command": "/issue:execute", "optional": false }
-          ]
-        }
-      },
-      "cli_hints": {
-        "discovery": { "tool": "gemini", "mode": "analysis", "trigger": "perspective_analysis", "parallel": true },
-        "solution_generation": { "tool": "gemini", "mode": "analysis", "trigger": "always", "parallel": true },
-        "batch_execution": { "tool": "codex", "mode": "write", "trigger": "always" }
-      },
-      "estimated_time": "1-4 hours"
-    },
-    "lite-lite-lite": {
-      "name": "Ultra-Lite Multi-CLI",
-      "description": "零文件 + 自动CLI选择 + 语义描述 + 直接执行",
-      "complexity": ["low", "medium"],
-      "steps": [
-        { "phase": "clarify", "description": "需求澄清 (AskUser if needed)" },
-        { "phase": "auto-select", "description": "任务分析 → 自动选择CLI组合" },
-        { "phase": "multi-cli", "description": "并行多CLI分析" },
-        { "phase": "decision", "description": "展示结果 → AskUser决策" },
-        { "phase": "execute", "description": "直接执行 (无中间文件)" }
-      ],
-      "vs_multi_cli_plan": {
-        "artifacts": "None vs IMPL_PLAN.md + plan.json + synthesis.json",
-        "session": "Stateless vs Persistent",
-        "cli_selection": "Auto-select based on task analysis vs Config-driven",
-        "iteration": "Via AskUser vs Via rounds/synthesis",
-        "execution": "Direct vs Via lite-execute",
-        "best_for": "Quick fixes, simple features vs Complex multi-step implementations"
-      },
-      "cli_hints": {
-        "analysis": { "tool": "auto", "mode": "analysis", "parallel": true },
-        "execution": { "tool": "auto", "mode": "write" }
-      },
-      "estimated_time": "10-30 min"
-    },
     "multi-cli-plan": {
       "name": "Multi-CLI Collaborative Planning",
+      "level": "L2",
       "description": "ACE上下文 + 多CLI协作分析 + 迭代收敛 + 计划生成",
       "complexity": ["medium", "high"],
+      "artifacts": ".workflow/.multi-cli-plan/{session}/",
       "steps": [
         { "command": "/workflow:multi-cli-plan", "optional": false, "phases": [
           "context_gathering: ACE语义搜索",
@@ -210,28 +142,154 @@
         "discussion": { "tools": ["gemini", "codex", "claude"], "mode": "analysis", "parallel": true },
         "planning": { "tool": "gemini", "mode": "analysis" }
       },
-      "output": ".workflow/.multi-cli-plan/{session-id}/",
       "estimated_time": "30-90 min"
     },
+    "coupled": {
+      "name": "Standard Planning",
+      "level": "L3",
+      "description": "完整规划 + 验证 + 执行",
+      "complexity": ["medium", "high"],
+      "artifacts": ".workflow/active/{session}/",
+      "steps": [
+        { "command": "/workflow:plan", "optional": false },
+        { "command": "/workflow:action-plan-verify", "optional": false, "auto_continue": true },
+        { "command": "/workflow:execute", "optional": false },
+        { "command": "/workflow:review", "optional": true }
+      ],
+      "cli_hints": {
+        "pre_analysis": { "tool": "gemini", "mode": "analysis", "trigger": "always" },
+        "execution": { "tool": "codex", "mode": "write", "trigger": "always" }
+      },
+      "estimated_time": "2-4 hours"
+    },
+    "full": {
+      "name": "Full Exploration (Brainstorm)",
+      "level": "L4",
+      "description": "头脑风暴 + 规划 + 执行",
+      "complexity": ["high"],
+      "artifacts": ".workflow/active/{session}/.brainstorming/",
+      "steps": [
+        { "command": "/workflow:brainstorm:auto-parallel", "optional": false, "confirm_before": true },
+        { "command": "/workflow:plan", "optional": false },
+        { "command": "/workflow:action-plan-verify", "optional": true, "auto_continue": true },
+        { "command": "/workflow:execute", "optional": false }
+      ],
+      "cli_hints": {
+        "role_analysis": { "tool": "gemini", "mode": "analysis", "trigger": "always", "parallel": true },
+        "execution": { "tool": "codex", "mode": "write", "trigger": "task_count >= 3" }
+      },
+      "estimated_time": "1-3 hours"
+    },
+    "bugfix": {
+      "name": "Bug Fix",
+      "level": "L2",
+      "description": "智能诊断 + 修复 (5 phases)",
+      "complexity": ["low", "medium"],
+      "artifacts": ".workflow/.lite-fix/{bug-slug}-{date}/",
+      "variants": {
+        "standard": [{ "command": "/workflow:lite-fix", "optional": false }],
+        "hotfix": [{ "command": "/workflow:lite-fix --hotfix", "optional": false }]
+      },
+      "phases": [
+        "Phase 1: Bug Analysis & Diagnosis (severity pre-assessment)",
+        "Phase 2: Clarification (optional, AskUserQuestion)",
+        "Phase 3: Fix Planning (Low/Medium → Claude, High/Critical → cli-lite-planning-agent)",
+        "Phase 4: Confirmation & Selection",
+        "Phase 5: Execute (→ lite-execute --mode bugfix)"
+      ],
+      "cli_hints": {
+        "diagnosis": { "tool": "gemini", "mode": "analysis", "trigger": "always" },
+        "fix": { "tool": "codex", "mode": "write", "trigger": "severity >= medium" }
+      },
+      "estimated_time": "10-30 min"
+    },
+    "issue": {
+      "name": "Issue Lifecycle",
+      "level": "Supplementary",
+      "description": "发现积累 → 批量规划 → 队列优化 → 并行执行 (Main Workflow 补充机制)",
+      "complexity": ["medium", "high"],
+      "artifacts": ".workflow/.issues/",
+      "purpose": "Post-development continuous maintenance, maintain main branch stability",
+      "phases": {
+        "accumulation": {
+          "description": "项目迭代中持续发现和积累issue",
+          "commands": ["/issue:discover", "/issue:discover-by-prompt", "/issue:new"],
+          "trigger": "post-task, code-review, test-failure"
+        },
+        "resolution": {
+          "description": "集中规划和执行积累的issue",
+          "steps": [
+            { "command": "/issue:plan --all-pending", "optional": false },
+            { "command": "/issue:queue", "optional": false },
+            { "command": "/issue:execute", "optional": false }
+          ]
+        }
+      },
+      "worktree_support": {
+        "description": "可选的 worktree 隔离，保持主分支稳定",
+        "use_case": "主开发完成后的 issue 修复"
+      },
+      "cli_hints": {
+        "discovery": { "tool": "gemini", "mode": "analysis", "trigger": "perspective_analysis", "parallel": true },
+        "solution_generation": { "tool": "gemini", "mode": "analysis", "trigger": "always", "parallel": true },
+        "batch_execution": { "tool": "codex", "mode": "write", "trigger": "always" }
+      },
+      "estimated_time": "1-4 hours"
+    },
     "tdd": {
       "name": "Test-Driven Development",
-      "description": "TDD规划 + 执行 + 验证",
+      "level": "L3",
+      "description": "TDD规划 + 执行 + 验证 (6 phases)",
       "complexity": ["medium", "high"],
+      "artifacts": ".workflow/active/{session}/",
       "steps": [
         { "command": "/workflow:tdd-plan", "optional": false },
+        { "command": "/workflow:action-plan-verify", "optional": true, "auto_continue": true },
         { "command": "/workflow:execute", "optional": false },
         { "command": "/workflow:tdd-verify", "optional": false }
       ],
+      "tdd_structure": {
+        "description": "Each IMPL task contains complete internal Red-Green-Refactor cycle",
+        "meta": "tdd_workflow: true",
+        "flow_control": "implementation_approach contains 3 steps (red/green/refactor)"
+      },
       "cli_hints": {
         "test_strategy": { "tool": "gemini", "mode": "analysis", "trigger": "always" },
         "red_green_refactor": { "tool": "codex", "mode": "write", "trigger": "always" }
       },
       "estimated_time": "1-3 hours"
     },
+    "test-fix": {
+      "name": "Test Fix Generation",
+      "level": "L3",
+      "description": "测试修复生成 + 执行循环 (5 phases)",
+      "complexity": ["medium", "high"],
+      "artifacts": ".workflow/active/WFS-test-{session}/",
+      "dual_mode": {
+        "session_mode": { "input": "WFS-xxx", "context_source": "Source session summaries" },
+        "prompt_mode": { "input": "Text/file path", "context_source": "Direct codebase analysis" }
+      },
+      "steps": [
+        { "command": "/workflow:test-fix-gen", "optional": false },
+        { "command": "/workflow:test-cycle-execute", "optional": false }
+      ],
+      "task_structure": [
+        "IMPL-001.json (test understanding & generation)",
+        "IMPL-001.5-review.json (quality gate)",
+        "IMPL-002.json (test execution & fix cycle)"
+      ],
+      "cli_hints": {
+        "analysis": { "tool": "gemini", "mode": "analysis", "trigger": "always" },
+        "fix_cycle": { "tool": "codex", "mode": "write", "trigger": "pass_rate < 0.95" }
+      },
+      "estimated_time": "1-2 hours"
+    },
     "ui": {
       "name": "UI-First Development",
+      "level": "L3/L4",
       "description": "UI设计 + 规划 + 执行",
       "complexity": ["medium", "high"],
+      "artifacts": ".workflow/active/{session}/",
       "variants": {
         "explore": [
           { "command": "/workflow:ui-design:explore-auto", "optional": false },
@@ -250,8 +308,10 @@
     },
     "review-fix": {
       "name": "Review and Fix",
+      "level": "L3",
       "description": "多维审查 + 自动修复",
       "complexity": ["medium"],
+      "artifacts": ".workflow/active/{session}/review_report.md",
       "steps": [
         { "command": "/workflow:review-session-cycle", "optional": false },
         { "command": "/workflow:review-fix", "optional": true }
@@ -264,6 +324,7 @@
     },
     "docs": {
       "name": "Documentation",
+      "level": "L2",
       "description": "批量文档生成",
       "complexity": ["low", "medium"],
       "variants": {
@@ -278,8 +339,17 @@
   },
 
   "intent_rules": {
+    "_level_mapping": {
+      "description": "Intent → Level → Flow mapping guide",
+      "L1": ["lite-lite-lite"],
+      "L2": ["rapid", "bugfix", "multi-cli-plan", "docs"],
+      "L3": ["coupled", "tdd", "test-fix", "review-fix", "ui"],
+      "L4": ["full"],
+      "Supplementary": ["issue"]
+    },
     "bugfix": {
       "priority": 1,
+      "level": "L2",
       "variants": {
         "hotfix": {
           "patterns": ["hotfix", "urgent", "production", "critical", "emergency", "紧急", "生产环境", "线上"],
@@ -293,6 +363,7 @@
     },
     "issue_batch": {
       "priority": 2,
+      "level": "Supplementary",
       "patterns": {
         "batch": ["issues", "batch", "queue", "多个", "批量"],
         "action": ["fix", "resolve", "处理", "解决"]
@@ -302,11 +373,25 @@
     },
     "exploration": {
       "priority": 3,
+      "level": "L4",
       "patterns": ["不确定", "不知道", "explore", "研究", "分析一下", "怎么做", "what if", "探索"],
       "flow": "full"
     },
-    "ui_design": {
+    "multi_perspective": {
+      "priority": 3,
+      "level": "L2",
+      "patterns": ["多视角", "权衡", "比较方案", "cross-verify", "多CLI", "协作分析"],
+      "flow": "multi-cli-plan"
+    },
+    "quick_task": {
       "priority": 4,
+      "level": "L1",
+      "patterns": ["快速", "简单", "small", "quick", "simple", "trivial", "小改动"],
+      "flow": "lite-lite-lite"
+    },
+    "ui_design": {
+      "priority": 5,
+      "level": "L3/L4",
       "patterns": ["ui", "界面", "design", "设计", "component", "组件", "style", "样式", "layout", "布局"],
       "variants": {
         "imitate": { "triggers": ["参考", "模仿", "像", "类似"], "flow": "ui.imitate" },
@@ -314,17 +399,26 @@
       }
     },
     "tdd": {
-      "priority": 5,
+      "priority": 6,
+      "level": "L3",
       "patterns": ["tdd", "test-driven", "测试驱动", "先写测试", "test first"],
       "flow": "tdd"
     },
+    "test_fix": {
+      "priority": 7,
+      "level": "L3",
+      "patterns": ["测试失败", "test fail", "fix test", "test error", "pass rate", "coverage gap"],
+      "flow": "test-fix"
+    },
     "review": {
-      "priority": 6,
+      "priority": 8,
+      "level": "L3",
       "patterns": ["review", "审查", "检查代码", "code review", "质量检查"],
       "flow": "review-fix"
     },
     "documentation": {
-      "priority": 7,
+      "priority": 9,
+      "level": "L2",
       "patterns": ["文档", "documentation", "docs", "readme"],
       "variants": {
         "incremental": { "triggers": ["更新", "增量"], "flow": "docs.incremental" },
@@ -334,9 +428,9 @@
     "feature": {
       "priority": 99,
       "complexity_map": {
-        "high": "coupled",
-        "medium": "rapid",
-        "low": "rapid"
+        "high": { "level": "L3", "flow": "coupled" },
+        "medium": { "level": "L2", "flow": "rapid" },
+        "low": { "level": "L1", "flow": "lite-lite-lite" }
       }
     }
   },

.claude/skills/skill-generator/specs/cli-integration.md

@@ -304,28 +304,22 @@ async function runWithTool(tool, context) {
 ### 引用协议模板
 
 ```bash
-# 分析模式 - 必须引用 analysis-protocol.md
+# Analysis mode - use --rule to auto-load protocol and template (appended to prompt)
 ccw cli -p "
-RULES: $(cat ~/.claude/workflows/cli-templates/protocols/analysis-protocol.md)
-$(cat ~/.claude/workflows/cli-templates/prompts/analysis/02-analyze-code-patterns.txt)
-..." --tool gemini --mode analysis
+CONSTRAINTS: ...
+..." --tool gemini --mode analysis --rule analysis-code-patterns
 
-# 写入模式 - 必须引用 write-protocol.md
+# Write mode - use --rule to auto-load protocol and template (appended to prompt)
 ccw cli -p "
-RULES: $(cat ~/.claude/workflows/cli-templates/protocols/write-protocol.md)
-$(cat ~/.claude/workflows/cli-templates/prompts/development/02-implement-feature.txt)
-..." --tool codex --mode write
+CONSTRAINTS: ...
+..." --tool codex --mode write --rule development-feature
 ```
 
 ### 动态模板构建
 
 ```javascript
 function buildPrompt(config) {
-  const { purpose, task, mode, context, expected, template } = config;
-
-  const protocolPath = mode === 'write'
-    ? '~/.claude/workflows/cli-templates/protocols/write-protocol.md'
-    : '~/.claude/workflows/cli-templates/protocols/analysis-protocol.md';
+  const { purpose, task, mode, context, expected, constraints } = config;
 
   return `
 PURPOSE: ${purpose}
@@ -333,8 +327,8 @@ TASK: ${task.map(t => `• ${t}`).join('\n')}
 MODE: ${mode}
 CONTEXT: ${context}
 EXPECTED: ${expected}
-RULES: $(cat ${protocolPath}) $(cat ${template})
-`;
+CONSTRAINTS: ${constraints || ''}
+`; // Use --rule option to auto-append protocol + template
 }
 ```
 
@@ -435,11 +429,11 @@ CLI 调用 (Bash + ccw cli):
 - 相关任务使用 `--resume` 保持上下文
 - 独立任务不使用 `--resume`
 
-### 4. 提示词规范
+### 4. Prompt Specification
 
-- 始终使用 PURPOSE/TASK/MODE/CONTEXT/EXPECTED/RULES 结构
-- 必须引用协议模板（analysis-protocol 或 write-protocol）
-- 使用 `$(cat ...)` 动态加载模板
+- Always use PURPOSE/TASK/MODE/CONTEXT/EXPECTED/CONSTRAINTS structure
+- Use `--rule <template>` to auto-append protocol + template to prompt
+- Template name format: `category-function` (e.g., `analysis-code-patterns`)
 
 ### 5. 结果处理
 

.claude/workflows/cli-templates/prompts/workflow-skill-conflict-patterns.txt

@@ -76,8 +76,8 @@ TASK: • Extract conflicts from IMPL_PLAN and lessons • Group by type (archit
 MODE: analysis
 CONTEXT: @.workflow/.archives/*/IMPL_PLAN.md @.workflow/.archives/manifest.json
 EXPECTED: Conflict patterns with frequency and resolution
-RULES: $(cat ~/.claude/workflows/cli-templates/prompts/workflow/skill-aggregation.txt) | analysis=READ-ONLY
-" --tool gemini --cd .workflow/.archives
+CONSTRAINTS: analysis=READ-ONLY
+" --tool gemini --mode analysis --rule workflow-skill-aggregation --cd .workflow/.archives
 ```
 
 **Pattern Grouping**:

.claude/workflows/cli-templates/prompts/workflow-skill-lessons-learned.txt

@@ -72,8 +72,8 @@ TASK: • Group successes by functional domain • Categorize challenges by seve
 MODE: analysis
 CONTEXT: @.workflow/.archives/manifest.json
 EXPECTED: Aggregated lessons with frequency counts
-RULES: $(cat ~/.claude/workflows/cli-templates/prompts/workflow/skill-aggregation.txt) | analysis=READ-ONLY
-" --tool gemini --cd .workflow/.archives
+CONSTRAINTS: analysis=READ-ONLY
+" --tool gemini --mode analysis --rule workflow-skill-aggregation --cd .workflow/.archives
 ```
 
 **Severity Classification**:

.claude/workflows/cli-templates/schemas/issues-jsonl-schema.json

@@ -65,6 +65,35 @@
       "items": { "type": "string" },
       "description": "Files/modules affected"
     },
+    "feedback": {
+      "type": "array",
+      "description": "Execution feedback history (failures, clarifications, rejections) for planning phase reference",
+      "items": {
+        "type": "object",
+        "required": ["type", "stage", "content", "created_at"],
+        "properties": {
+          "type": {
+            "type": "string",
+            "enum": ["failure", "clarification", "rejection"],
+            "description": "Type of feedback"
+          },
+          "stage": {
+            "type": "string",
+            "enum": ["new", "plan", "execute"],
+            "description": "Which stage the feedback occurred (new=creation, plan=planning, execute=execution)"
+          },
+          "content": {
+            "type": "string",
+            "description": "JSON string for failures (with solution_id, task_id, error_type, message, stack_trace) or plain text for clarifications/rejections"
+          },
+          "created_at": {
+            "type": "string",
+            "format": "date-time",
+            "description": "Timestamp when feedback was created"
+          }
+        }
+      }
+    },
     "lifecycle_requirements": {
       "type": "object",
       "properties": {

.claude/workflows/cli-templates/schemas/plan-json-schema.json

@@ -143,11 +143,211 @@
               }
             },
             "description": "CLI execution strategy based on task dependencies"
+          },
+          "rationale": {
+            "type": "object",
+            "properties": {
+              "chosen_approach": {
+                "type": "string",
+                "description": "The selected implementation approach and why it was chosen"
+              },
+              "alternatives_considered": {
+                "type": "array",
+                "items": {"type": "string"},
+                "description": "Alternative approaches that were considered but not chosen"
+              },
+              "decision_factors": {
+                "type": "array",
+                "items": {"type": "string"},
+                "description": "Key factors that influenced the decision (performance, maintainability, cost, etc.)"
+              },
+              "tradeoffs": {
+                "type": "string",
+                "description": "Known tradeoffs of the chosen approach"
+              }
+            },
+            "description": "Design rationale explaining WHY this approach was chosen (required for Medium/High complexity)"
+          },
+          "verification": {
+            "type": "object",
+            "properties": {
+              "unit_tests": {
+                "type": "array",
+                "items": {"type": "string"},
+                "description": "List of unit test names/descriptions to create"
+              },
+              "integration_tests": {
+                "type": "array",
+                "items": {"type": "string"},
+                "description": "List of integration test names/descriptions to create"
+              },
+              "manual_checks": {
+                "type": "array",
+                "items": {"type": "string"},
+                "description": "Manual verification steps with specific actions"
+              },
+              "success_metrics": {
+                "type": "array",
+                "items": {"type": "string"},
+                "description": "Quantified metrics for success (e.g., 'Response time <200ms', 'Coverage >80%')"
+              }
+            },
+            "description": "Detailed verification steps beyond acceptance criteria (required for Medium/High complexity)"
+          },
+          "risks": {
+            "type": "array",
+            "items": {
+              "type": "object",
+              "required": ["description", "probability", "impact", "mitigation"],
+              "properties": {
+                "description": {
+                  "type": "string",
+                  "description": "Description of the risk"
+                },
+                "probability": {
+                  "type": "string",
+                  "enum": ["Low", "Medium", "High"],
+                  "description": "Likelihood of the risk occurring"
+                },
+                "impact": {
+                  "type": "string",
+                  "enum": ["Low", "Medium", "High"],
+                  "description": "Impact severity if the risk occurs"
+                },
+                "mitigation": {
+                  "type": "string",
+                  "description": "Strategy to mitigate or prevent the risk"
+                },
+                "fallback": {
+                  "type": "string",
+                  "description": "Alternative approach if mitigation fails"
+                }
+              }
+            },
+            "description": "Risk assessment and mitigation strategies (required for High complexity)"
+          },
+          "code_skeleton": {
+            "type": "object",
+            "properties": {
+              "interfaces": {
+                "type": "array",
+                "items": {
+                  "type": "object",
+                  "properties": {
+                    "name": {"type": "string"},
+                    "definition": {"type": "string"},
+                    "purpose": {"type": "string"}
+                  }
+                },
+                "description": "Key interface/type definitions"
+              },
+              "key_functions": {
+                "type": "array",
+                "items": {
+                  "type": "object",
+                  "properties": {
+                    "signature": {"type": "string"},
+                    "purpose": {"type": "string"},
+                    "returns": {"type": "string"}
+                  }
+                },
+                "description": "Critical function signatures"
+              },
+              "classes": {
+                "type": "array",
+                "items": {
+                  "type": "object",
+                  "properties": {
+                    "name": {"type": "string"},
+                    "purpose": {"type": "string"},
+                    "methods": {
+                      "type": "array",
+                      "items": {"type": "string"}
+                    }
+                  }
+                },
+                "description": "Key class structures"
+              }
+            },
+            "description": "Code skeleton with interface/function signatures (required for High complexity)"
           }
         }
       },
       "description": "Structured task breakdown (1-10 tasks)"
     },
+    "data_flow": {
+      "type": "object",
+      "properties": {
+        "diagram": {
+          "type": "string",
+          "description": "ASCII/text representation of data flow (e.g., 'A → B → C')"
+        },
+        "stages": {
+          "type": "array",
+          "items": {
+            "type": "object",
+            "required": ["stage", "input", "output", "component"],
+            "properties": {
+              "stage": {
+                "type": "string",
+                "description": "Stage name (e.g., 'Extraction', 'Processing', 'Storage')"
+              },
+              "input": {
+                "type": "string",
+                "description": "Input data format/type"
+              },
+              "output": {
+                "type": "string",
+                "description": "Output data format/type"
+              },
+              "component": {
+                "type": "string",
+                "description": "Component/module handling this stage"
+              },
+              "transformations": {
+                "type": "array",
+                "items": {"type": "string"},
+                "description": "Data transformations applied in this stage"
+              }
+            }
+          },
+          "description": "Detailed data flow stages"
+        },
+        "dependencies": {
+          "type": "array",
+          "items": {"type": "string"},
+          "description": "External dependencies or data sources"
+        }
+      },
+      "description": "Global data flow design showing how data moves through the system (required for High complexity)"
+    },
+    "design_decisions": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "required": ["decision", "rationale"],
+        "properties": {
+          "decision": {
+            "type": "string",
+            "description": "The design decision made"
+          },
+          "rationale": {
+            "type": "string",
+            "description": "Why this decision was made"
+          },
+          "tradeoff": {
+            "type": "string",
+            "description": "What was traded off for this decision"
+          },
+          "alternatives": {
+            "type": "array",
+            "items": {"type": "string"},
+            "description": "Alternatives that were considered"
+          }
+        }
+      },
+      "description": "Global design decisions that affect the entire plan"
+    },
     "flow_control": {
       "type": "object",
       "properties": {

.claude/workflows/cli-tools-usage.md

@@ -5,7 +5,7 @@
 2. [Tool Selection](#tool-selection)
 3. [Prompt Template](#prompt-template)
 4. [CLI Execution](#cli-execution)
-5. [Execution Configuration](#execution-configuration)
+5. [Auto-Invoke Triggers](#auto-invoke-triggers)
 6. [Best Practices](#best-practices)
 
 ---
@@ -110,13 +110,13 @@ When primary tool fails or is unavailable:
 
 ### Universal Prompt Template
 
-```
-PURPOSE: [what] + [why] + [success criteria] + [constraints/scope]
+```bash
+ccw cli -p "PURPOSE: [what] + [why] + [success criteria] + [constraints/scope]
 TASK: • [step 1: specific action] • [step 2: specific action] • [step 3: specific action]
 MODE: [analysis|write]
 CONTEXT: @[file patterns] | Memory: [session/tech/module context]
 EXPECTED: [deliverable format] + [quality criteria] + [structure requirements]
-RULES: $(cat ~/.claude/workflows/cli-templates/protocols/[mode]-protocol.md) $(cat ~/.claude/workflows/cli-templates/prompts/[category]/[template].txt) | [domain constraints]
+CONSTRAINTS: [domain constraints]" --tool <tool-id> --mode <analysis|write> --rule <category-template>
 ```
 
 ### Intent Capture Checklist (Before CLI Execution)
@@ -165,11 +165,11 @@ Every command MUST include these fields:
   - Bad Example: "Report"
   - Good Example: "Markdown report with: severity levels (Critical/High/Medium/Low), file:line references, remediation code snippets, priority ranking"
 
-- **RULES**
-  - Purpose: Protocol + template + constraints
-  - Components: $(cat protocol) + $(cat template) + domain rules
-  - Bad Example: (missing)
-  - Good Example: "$(cat ~/.claude/workflows/cli-templates/protocols/analysis-protocol.md) $(cat ~/.claude/workflows/cli-templates/prompts/analysis/03-assess-security-risks.txt) \| Focus on authentication \| Ignore test files"
+- **CONSTRAINTS**
+  - Purpose: Domain-specific constraints
+  - Components: Scope limits, special requirements, focus areas
+  - Bad Example: (missing or too vague)
+  - Good Example: "Focus on authentication | Ignore test files | No breaking changes"
 
 ### CONTEXT Configuration
 
@@ -204,7 +204,11 @@ Memory: Integration with auth module, using shared error patterns from @shared/u
 For complex requirements, discover files BEFORE CLI execution:
 
 ```bash
-# Step 1: Discover files
+# Step 1: Discover files (choose one method)
+# Method A: ACE semantic search (recommended)
+mcp__ace-tool__search_context(project_root_path="/path", query="React components with export")
+
+# Method B: Ripgrep pattern search
 rg "export.*Component" --files-with-matches --type ts
 
 # Step 2: Build CONTEXT
@@ -214,108 +218,60 @@ CONTEXT: @components/Auth.tsx @types/auth.d.ts | Memory: Previous type refactori
 ccw cli -p "..." --tool <tool-id> --mode analysis --cd src
 ```
 
-### RULES Configuration
+### --rule Configuration
 
-**Format**: `RULES: $(cat ~/.claude/workflows/cli-templates/prompts/[category]/[template].txt) | [constraints]`
+**Use `--rule` option to auto-load templates**:
 
-**⚠️ MANDATORY**: Exactly ONE template reference is REQUIRED. Select from Task-Template Matrix or use universal fallback:
-- `universal/00-universal-rigorous-style.txt` - For precision-critical tasks (default fallback)
-- `universal/00-universal-creative-style.txt` - For exploratory tasks
-
-**Command Substitution Rules**:
-- Use `$(cat ...)` directly in **double quotes** - command substitution executes in your local shell BEFORE passing to ccw
-- Shell expands `$(cat ...)` into file content automatically - do NOT read template content first
-- NEVER use escape characters (`\$`, `\"`, `\'`) or single quotes - these prevent shell expansion
-- Tilde (`~`) expands correctly in prompt context
-
-**Critical**: Use double quotes `"..."` around the entire prompt to enable `$(cat ...)` expansion:
 ```bash
-# ✓ CORRECT - double quotes allow shell expansion
-ccw cli -p "RULES: $(cat ~/.claude/workflows/cli-templates/protocols/analysis-protocol.md) ..." --tool <tool-id>
-
-# ✗ WRONG - single quotes prevent expansion
-ccw cli -p 'RULES: $(cat ~/.claude/workflows/cli-templates/protocols/analysis-protocol.md) ...' --tool <tool-id>
-
-# ✗ WRONG - escaped $ prevents expansion
-ccw cli -p "RULES: \$(cat ~/.claude/workflows/cli-templates/protocols/analysis-protocol.md) ..." --tool <tool-id>
+ccw cli -p "..." --tool gemini --mode analysis --rule analysis-review-architecture
 ```
 
-### Mode Protocol References (MANDATORY)
+### Mode Protocol References
 
-**⚠️ REQUIRED**: Every CLI execution MUST include the corresponding mode protocol in RULES:
-
-#### Mode Rule Templates
-
-**Purpose**: Mode protocols define permission boundaries and operational constraints for each execution mode.
+**`--rule` auto-loads Protocol based on mode**:
+- `--mode analysis` → analysis-protocol.md
+- `--mode write` → write-protocol.md
 
 **Protocol Mapping**:
 
 - **`analysis`** mode
-  - Protocol: `$(cat ~/.claude/workflows/cli-templates/protocols/analysis-protocol.md)`
-  - Permission: Read-only operations
-  - Enforces: No file creation/modification/deletion
+  - Permission: Read-only
+  - Constraint: No file create/modify/delete
 
 - **`write`** mode
-  - Protocol: `$(cat ~/.claude/workflows/cli-templates/protocols/write-protocol.md)`
   - Permission: Create/Modify/Delete files
-  - Enforces: Explicit write authorization and full workflow execution capability
-
-**RULES Format** (protocol MUST be included):
-```bash
-# Analysis mode - MUST include analysis-protocol.md
-RULES: $(cat ~/.claude/workflows/cli-templates/protocols/analysis-protocol.md) $(cat ~/.claude/workflows/cli-templates/prompts/analysis/...) | constraints
-
-# Write mode - MUST include write-protocol.md
-RULES: $(cat ~/.claude/workflows/cli-templates/protocols/write-protocol.md) $(cat ~/.claude/workflows/cli-templates/prompts/development/...) | constraints
-```
-
-**Validation**: CLI execution without mode protocol reference is INVALID
-
-**Why Mode Rules Are Required**:
-- Ensures consistent permission enforcement across all tools
-- Prevents accidental file modifications during analysis tasks
-- Provides explicit authorization trail for write operations
-- Enables safe automation with clear boundaries
+  - Constraint: Full workflow execution
 
 ### Template System
 
-**Base Path**: `~/.claude/workflows/cli-templates/prompts/`
+**Available `--rule` template names**:
 
-**Naming Convention**:
-- `00-*` - Universal fallbacks (when no specific match)
-- `01-*` - Universal, high-frequency
-- `02-*` - Common specialized
-- `03-*` - Domain-specific
-
-**Universal Templates**:
-
-- **`universal/00-universal-rigorous-style.txt`**: Precision-critical, systematic methodology
-- **`universal/00-universal-creative-style.txt`**: Exploratory, innovative solutions
-
-**Task-Template Matrix**:
+**Universal**:
+- `universal-rigorous-style` - Precise tasks
+- `universal-creative-style` - Exploratory tasks
 
 **Analysis**:
-- Execution Tracing: `analysis/01-trace-code-execution.txt`
-- Bug Diagnosis: `analysis/01-diagnose-bug-root-cause.txt`
-- Code Patterns: `analysis/02-analyze-code-patterns.txt`
-- Document Analysis: `analysis/02-analyze-technical-document.txt`
-- Architecture Review: `analysis/02-review-architecture.txt`
-- Code Review: `analysis/02-review-code-quality.txt`
-- Performance: `analysis/03-analyze-performance.txt`
-- Security: `analysis/03-assess-security-risks.txt`
+- `analysis-trace-code-execution` - Execution tracing
+- `analysis-diagnose-bug-root-cause` - Bug diagnosis
+- `analysis-analyze-code-patterns` - Code patterns
+- `analysis-analyze-technical-document` - Document analysis
+- `analysis-review-architecture` - Architecture review
+- `analysis-review-code-quality` - Code review
+- `analysis-analyze-performance` - Performance analysis
+- `analysis-assess-security-risks` - Security assessment
 
 **Planning**:
-- Architecture: `planning/01-plan-architecture-design.txt`
-- Task Breakdown: `planning/02-breakdown-task-steps.txt`
-- Component Design: `planning/02-design-component-spec.txt`
-- Migration: `planning/03-plan-migration-strategy.txt`
+- `planning-plan-architecture-design` - Architecture design
+- `planning-breakdown-task-steps` - Task breakdown
+- `planning-design-component-spec` - Component design
+- `planning-plan-migration-strategy` - Migration strategy
 
 **Development**:
-- Feature: `development/02-implement-feature.txt`
-- Refactoring: `development/02-refactor-codebase.txt`
-- Tests: `development/02-generate-tests.txt`
-- UI Component: `development/02-implement-component-ui.txt`
-- Debugging: `development/03-debug-runtime-issues.txt`
+- `development-implement-feature` - Feature implementation
+- `development-refactor-codebase` - Code refactoring
+- `development-generate-tests` - Test generation
+- `development-implement-component-ui` - UI component
+- `development-debug-runtime-issues` - Runtime debugging
 
 ---
 
@@ -336,10 +292,13 @@ RULES: $(cat ~/.claude/workflows/cli-templates/protocols/write-protocol.md) $(ca
 - **`review`**
   - Permission: Read-only (code review output)
   - Use For: Git-aware code review of uncommitted changes, branch diffs, specific commits
-  - Specification: **codex only** - uses `codex review` subcommand with `--uncommitted` by default
+  - Specification: **codex only** - uses `codex review` subcommand
   - Tool Behavior:
-    - `codex`: Executes `codex review --uncommitted [prompt]` for structured code review
+    - `codex`: Executes `codex review` for structured code review
     - Other tools (gemini/qwen/claude): Accept mode but no operation change (treated as analysis)
+  - **Constraint**: Target flags (`--uncommitted`, `--base`, `--commit`) and prompt are mutually exclusive
+    - With prompt only: `ccw cli -p "Focus on security" --tool codex --mode review` (reviews uncommitted by default)
+    - With target flag only: `ccw cli --tool codex --mode review --commit abc123` (no prompt allowed)
 
 ### Command Options
 
@@ -368,6 +327,11 @@ RULES: $(cat ~/.claude/workflows/cli-templates/protocols/write-protocol.md) $(ca
   - Description: Resume previous session
   - Default: -
 
+- **`--rule <template>`**
+  - Description: Template name, auto-loads protocol + template appended to prompt
+  - Default: universal-rigorous-style
+  - Auto-selects protocol based on --mode
+
 ### Directory Configuration
 
 #### Working Directory (`--cd`)
@@ -429,62 +393,60 @@ ASSISTANT RESPONSE: [Previous output]
 
 **Analysis Task** (Security Audit):
 ```bash
-ccw cli -p "
-PURPOSE: Identify OWASP Top 10 vulnerabilities in authentication module to pass security audit; success = all critical/high issues documented with remediation
+ccw cli -p "PURPOSE: Identify OWASP Top 10 vulnerabilities in authentication module to pass security audit; success = all critical/high issues documented with remediation
 TASK: • Scan for injection flaws (SQL, command, LDAP) • Check authentication bypass vectors • Evaluate session management • Assess sensitive data exposure
 MODE: analysis
 CONTEXT: @src/auth/**/* @src/middleware/auth.ts | Memory: Using bcrypt for passwords, JWT for sessions
 EXPECTED: Security report with: severity matrix, file:line references, CVE mappings where applicable, remediation code snippets prioritized by risk
-RULES: $(cat ~/.claude/workflows/cli-templates/protocols/analysis-protocol.md) $(cat ~/.claude/workflows/cli-templates/prompts/analysis/03-assess-security-risks.txt) | Focus on authentication | Ignore test files
-" --tool <tool-id> --mode analysis --cd src/auth
+CONSTRAINTS: Focus on authentication | Ignore test files
+" --tool gemini --mode analysis --rule analysis-assess-security-risks --cd src/auth
 ```
 
 **Implementation Task** (New Feature):
 ```bash
-ccw cli -p "
-PURPOSE: Implement rate limiting for API endpoints to prevent abuse; must be configurable per-endpoint; backward compatible with existing clients
+ccw cli -p "PURPOSE: Implement rate limiting for API endpoints to prevent abuse; must be configurable per-endpoint; backward compatible with existing clients
 TASK: • Create rate limiter middleware with sliding window • Implement per-route configuration • Add Redis backend for distributed state • Include bypass for internal services
 MODE: write
 CONTEXT: @src/middleware/**/* @src/config/**/* | Memory: Using Express.js, Redis already configured, existing middleware pattern in auth.ts
 EXPECTED: Production-ready code with: TypeScript types, unit tests, integration test, configuration example, migration guide
-RULES: $(cat ~/.claude/workflows/cli-templates/protocols/write-protocol.md) $(cat ~/.claude/workflows/cli-templates/prompts/development/02-implement-feature.txt) | Follow existing middleware patterns | No breaking changes
-" --tool <tool-id> --mode write
+CONSTRAINTS: Follow existing middleware patterns | No breaking changes
+" --tool gemini --mode write --rule development-implement-feature
 ```
 
 **Bug Fix Task**:
 ```bash
-ccw cli -p "
-PURPOSE: Fix memory leak in WebSocket connection handler causing server OOM after 24h; root cause must be identified before any fix
+ccw cli -p "PURPOSE: Fix memory leak in WebSocket connection handler causing server OOM after 24h; root cause must be identified before any fix
 TASK: • Trace connection lifecycle from open to close • Identify event listener accumulation • Check cleanup on disconnect • Verify garbage collection eligibility
 MODE: analysis
 CONTEXT: @src/websocket/**/* @src/services/connection-manager.ts | Memory: Using ws library, ~5000 concurrent connections in production
 EXPECTED: Root cause analysis with: memory profile, leak source (file:line), fix recommendation with code, verification steps
-RULES: $(cat ~/.claude/workflows/cli-templates/protocols/analysis-protocol.md) $(cat ~/.claude/workflows/cli-templates/prompts/analysis/01-diagnose-bug-root-cause.txt) | Focus on resource cleanup
-" --tool <tool-id> --mode analysis --cd src
+CONSTRAINTS: Focus on resource cleanup
+" --tool gemini --mode analysis --rule analysis-diagnose-bug-root-cause --cd src
 ```
 
 **Refactoring Task**:
 ```bash
-ccw cli -p "
-PURPOSE: Refactor payment processing to use strategy pattern for multi-gateway support; no functional changes; all existing tests must pass
+ccw cli -p "PURPOSE: Refactor payment processing to use strategy pattern for multi-gateway support; no functional changes; all existing tests must pass
 TASK: • Extract gateway interface from current implementation • Create strategy classes for Stripe, PayPal • Implement factory for gateway selection • Migrate existing code to use strategies
 MODE: write
 CONTEXT: @src/payments/**/* @src/types/payment.ts | Memory: Currently only Stripe, adding PayPal next sprint, must support future gateways
 EXPECTED: Refactored code with: strategy interface, concrete implementations, factory class, updated tests, migration checklist
-RULES: $(cat ~/.claude/workflows/cli-templates/protocols/write-protocol.md) $(cat ~/.claude/workflows/cli-templates/prompts/development/02-refactor-codebase.txt) | Preserve all existing behavior | Tests must pass
-" --tool <tool-id> --mode write
+CONSTRAINTS: Preserve all existing behavior | Tests must pass
+" --tool gemini --mode write --rule development-refactor-codebase
 ```
 
 **Code Review Task** (codex review mode):
 ```bash
-# Review uncommitted changes (default)
+# Option 1: Custom prompt (reviews uncommitted changes by default)
 ccw cli -p "Focus on security vulnerabilities and error handling" --tool codex --mode review
 
-# Review with custom instructions
-ccw cli -p "Check for breaking changes in API contracts and backward compatibility" --tool codex --mode review
+# Option 2: Target flag only (no prompt allowed with target flags)
+ccw cli --tool codex --mode review --uncommitted
+ccw cli --tool codex --mode review --base main
+ccw cli --tool codex --mode review --commit abc123
 ```
 
-> **Note**: `--mode review` only triggers special behavior for `codex` tool (uses `codex review --uncommitted`). Other tools accept the mode but execute as standard analysis.
+> **Note**: `--mode review` only triggers special behavior for `codex` tool. Target flags (`--uncommitted`, `--base`, `--commit`) and prompt are **mutually exclusive** - use one or the other, not both.
 
 ---
 
@@ -500,6 +462,41 @@ ccw cli -p "Check for breaking changes in API contracts and backward compatibili
 
 ---
 
+## Auto-Invoke Triggers
+
+**Proactive CLI invocation** - Auto-invoke `ccw cli` when encountering these scenarios:
+
+| Trigger Condition | Suggested Rule | When to Use |
+|-------------------|----------------|-------------|
+| **Self-repair fails** | `analysis-diagnose-bug-root-cause` | After 1+ failed fix attempts |
+| **Ambiguous requirements** | `planning-breakdown-task-steps` | Task description lacks clarity |
+| **Architecture decisions** | `planning-plan-architecture-design` | Complex feature needs design |
+| **Pattern uncertainty** | `analysis-analyze-code-patterns` | Unsure of existing conventions |
+| **Critical code paths** | `analysis-assess-security-risks` | Security/performance sensitive |
+
+### Execution Principles
+
+- **Default mode**: `--mode analysis` (read-only, safe for auto-execution)
+- **No confirmation needed**: Invoke proactively when triggers match
+- **Wait for results**: Complete analysis before next action
+- **Tool selection**: Use context-appropriate tool or fallback chain (`gemini` → `qwen` → `codex`)
+- **Rule flexibility**: Suggested rules are guidelines, not requirements - choose the most appropriate template for the situation
+
+### Example: Bug Fix with Auto-Invoke
+
+```bash
+# After 1+ failed fix attempts, auto-invoke root cause analysis
+ccw cli -p "PURPOSE: Identify root cause of [bug description]; success = actionable fix strategy
+TASK: • Trace execution flow • Identify failure point • Analyze state at failure • Determine fix approach
+MODE: analysis
+CONTEXT: @src/module/**/* | Memory: Previous fix attempts failed at [location]
+EXPECTED: Root cause analysis with: failure mechanism, stack trace interpretation, fix recommendation with code
+CONSTRAINTS: Focus on [specific area]
+" --tool gemini --mode analysis --rule analysis-diagnose-bug-root-cause
+```
+
+---
+
 ## Best Practices
 
 ### Core Principles
@@ -509,16 +506,15 @@ ccw cli -p "Check for breaking changes in API contracts and backward compatibili
 - **Use tools early and often** - Tools are faster and more thorough
 - **Unified CLI** - Always use `ccw cli -p` for consistent parameter handling
 - **Default mode is analysis** - Omit `--mode` for read-only operations, explicitly use `--mode write` for file modifications
-- **One template required** - ALWAYS reference exactly ONE template in RULES (use universal fallback if no specific match)
+- **Use `--rule` for templates** - Auto-loads protocol + template appended to prompt
 - **Write protection** - Require EXPLICIT `--mode write` for file operations
-- **Use double quotes for shell expansion** - Always wrap prompts in double quotes `"..."` to enable `$(cat ...)` command substitution; NEVER use single quotes or escape characters (`\$`, `\"`, `\'`)
 
 ### Workflow Principles
 
 - **Use CCW unified interface** for all executions
-- **Always include template** - Use Task-Template Matrix or universal fallback
+- **Always include template** - Use `--rule <template-name>` to load templates
 - **Be specific** - Clear PURPOSE, TASK, EXPECTED fields
-- **Include constraints** - File patterns, scope in RULES
+- **Include constraints** - File patterns, scope in CONSTRAINTS
 - **Leverage memory context** when building on previous work
 - **Discover patterns first** - Use rg/MCP before CLI execution
 - **Default to full context** - Use `@**/*` unless specific files needed
@@ -530,8 +526,8 @@ ccw cli -p "Check for breaking changes in API contracts and backward compatibili
 - [ ] **Context gathered** - File references + memory (default `@**/*`)
 - [ ] **Directory navigation** - `--cd` and/or `--includeDirs`
 - [ ] **Tool selected** - Explicit `--tool` or tag-based auto-selection
-- [ ] **Template applied (REQUIRED)** - Use specific or universal fallback template
-- [ ] **Constraints specified** - Scope, requirements
+- [ ] **Rule template** - `--rule <template-name>` loads template
+- [ ] **Constraints** - Domain constraints in CONSTRAINTS field
 
 ### Execution Workflow