fix: Correct TodoWrite Template in workflow execute command

- Replace markdown template with Claude Code TodoWrite tool usage - Update documentation to use built-in TodoWrite API instead of manual TODO_LIST.md updates - Align with JSON-only data model and real-time progress tracking principles - Add proper TodoWrite integration rules and examples 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>
2026-02-13 02:41:50 +08:00 · 2025-09-23 20:53:03 +08:00
parent 0f01cecc2d
commit a9555f2fd5
15 changed files with 1317 additions and 1081 deletions
--- a/.claude/commands/workflow/brainstorm/auto.md
+++ b/.claude/commands/workflow/brainstorm/auto.md
@@ -0,0 +1,217 @@
+---
+name: auto
+description: Intelligent brainstorming automation with dynamic role selection and guided context gathering
+usage: /workflow:brainstorm:auto "<topic>"
+argument-hint: "topic or challenge description"
+examples:
+  - /workflow:brainstorm:auto "Build real-time collaboration feature"
+  - /workflow:brainstorm:auto "Optimize database performance for millions of users"
+  - /workflow:brainstorm:auto "Implement secure authentication system"
+allowed-tools: Task(*), TodoWrite(*), Read(*), Write(*), Bash(*), Glob(*)
+---
+
+# Workflow Brainstorm Auto Command
+
+## Usage
+```bash
+/workflow:brainstorm:auto "<topic>"
+```
+
+## Role Selection Logic
+- **Technical & Architecture**: `architecture|system|performance|database|security` → system-architect, data-architect, security-expert
+- **Product & UX**: `user|ui|ux|interface|design|product|feature` → ui-designer, user-researcher, product-manager
+- **Business & Process**: `business|process|workflow|cost|innovation|testing` → business-analyst, innovation-lead, test-strategist
+- **Multi-role**: Complex topics automatically select 2-3 complementary roles
+- **Default**: `product-manager` if no clear match
+
+**Template Loading**: `~/.claude/scripts/planning-role-load.sh load <role-name>`
+**Template Source**: `.claude/workflows/cli-templates/planning-roles/`
+**Available Roles**: business-analyst, data-architect, feature-planner, innovation-lead, product-manager, security-expert, system-architect, test-strategist, ui-designer, user-researcher
+
+**Example**:
+```bash
+~/.claude/scripts/planning-role-load.sh load system-architect
+~/.claude/scripts/planning-role-load.sh load ui-designer
+~/.claude/scripts/planning-role-load.sh list  # Show all available roles
+```
+
+## Core Workflow
+
+### Analysis & Planning Process
+The command performs dedicated role analysis through:
+
+**0. Session Management** ⚠️ FIRST STEP
+- **Active session detection**: Check `.workflow/.active-*` markers
+- **Session selection**: Prompt user if multiple active sessions found
+- **Auto-creation**: Create `WFS-[topic-slug]` only if no active session exists
+- **Context isolation**: Each session maintains independent brainstorming state
+
+**1. Role Selection & Template Loading**
+- **Keyword analysis**: Extract topic keywords and map to planning roles
+- **Template loading**: Load role templates via `~/.claude/scripts/planning-role-load.sh`
+- **Role validation**: Verify against `.claude/workflows/cli-templates/planning-roles/`
+- **Multi-role detection**: Select 1-3 complementary roles based on topic complexity
+
+**2. Sequential Role Processing** ⚠️ CRITICAL ARCHITECTURE
+- **One Role = One Agent**: Each role gets dedicated conceptual-planning-agent
+- **Context gathering**: Role-specific questioning with validation
+- **Agent submission**: Complete context handoff to single-role agents
+- **Progress tracking**: Real-time TodoWrite updates per role
+
+**3. Analysis Artifacts Generated**
+- **Role contexts**: `.brainstorming/[role]-context.md` - User responses per role
+- **Agent outputs**: `.brainstorming/[role]/analysis.md` - Dedicated role analysis
+- **Session metadata**: `.brainstorming/auto-session.json` - Agent assignments and validation
+- **Synthesis**: `.brainstorming/synthesis/integrated-analysis.md` - Multi-role integration
+
+## Implementation Standards
+
+### Dedicated Agent Architecture ⚠️ CRITICAL
+Agents receive dedicated role assignments with complete context isolation:
+
+```json
+"agent_assignment": {
+  "role": "system-architect",
+  "agent_id": "conceptual-planning-agent-system-architect",
+  "context_source": ".brainstorming/system-architect-context.md",
+  "output_location": ".brainstorming/system-architect/",
+  "flow_control": {
+    "pre_analysis": [
+      {
+        "step": "load_role_template",
+        "action": "Load system-architect planning template",
+        "command": "bash(~/.claude/scripts/planning-role-load.sh load system-architect)",
+        "output_to": "role_template"
+      }
+    ],
+    "implementation_approach": {
+      "task_description": "Execute dedicated system-architect conceptual analysis for: [topic]",
+      "role_focus": "system-architect",
+      "user_context": "Direct user responses from context gathering phase",
+      "deliverables": "conceptual_analysis, strategic_recommendations, role_perspective"
+    }
+  }
+}
+```
+
+**Context Accumulation & Role Isolation**:
+1. **Role template loading**: Planning role template with domain expertise via CLI
+2. **Context validation**: Minimum response requirements with re-prompting
+3. **Conceptual analysis**: Role-specific perspective on topic without implementation details
+4. **Agent delegation**: Complete context handoff to dedicated conceptual-planning-agent
+
+**Content Sources**:
+- Role templates: `~/.claude/scripts/planning-role-load.sh load <role>` from `.claude/workflows/cli-templates/planning-roles/`
+- User responses: Direct context gathering during interactive questioning phase
+- Conceptual focus: Strategic and planning perspective without technical implementation
+
+**Trigger Conditions**: Topic analysis matches role domains, user provides adequate context responses, role template successfully loaded
+
+### Role Processing Standards
+
+**Core Principles**:
+1. **Sequential Processing** - Complete each role fully before proceeding to next
+2. **Context Validation** - Ensure adequate detail before agent submission
+3. **Dedicated Assignment** - One conceptual-planning-agent per role
+4. **Progress Tracking** - Real-time TodoWrite updates for role processing stages
+
+**Implementation Rules**:
+- **Maximum 3 roles**: Auto-selected based on topic complexity and domain overlap
+- **Context validation**: Minimum response length and completeness checks
+- **Agent isolation**: Each agent receives only role-specific context
+- **Error recovery**: Role-specific validation and retry logic
+
+**Role Question Templates**:
+- **system-architect**: Scale requirements, integration needs, technology constraints, non-functional requirements
+- **security-expert**: Sensitive data types, compliance requirements, threat concerns, auth/authz needs
+- **ui-designer**: User personas, platform support, design guidelines, accessibility requirements
+- **product-manager**: Business objectives, stakeholders, success metrics, timeline constraints
+- **data-architect**: Data types, volume projections, compliance needs, analytics requirements
+
+### Session Management ⚠️ CRITICAL
+- **⚡ FIRST ACTION**: Check for all `.workflow/.active-*` markers before role processing
+- **Multiple sessions support**: Different Claude instances can have different active brainstorming sessions
+- **User selection**: If multiple active sessions found, prompt user to select which one to work with
+- **Auto-session creation**: `WFS-[topic-slug]` only if no active session exists
+- **Session continuity**: MUST use selected active session for all role processing
+- **Context preservation**: Each role's context and agent output stored in session directory
+- **Session isolation**: Each session maintains independent brainstorming state and role assignments
+
+## Document Generation
+
+**Workflow**: Role Selection → Context Gathering → Agent Delegation → Documentation → Synthesis
+
+**Always Created**:
+- **auto-session.json**: Agent assignments, context validation, completion tracking
+- **[role]-context.md**: User responses per role with question-answer pairs
+
+**Auto-Created (per role)**:
+- **[role]/analysis.md**: Main role analysis from dedicated agent
+- **[role]/recommendations.md**: Role-specific recommendations
+- **[role]-template.md**: Loaded role planning template
+
+**Auto-Created (multi-role)**:
+- **synthesis/integrated-analysis.md**: Cross-role integration and consensus analysis
+- **synthesis/consensus-matrix.md**: Agreement/disagreement analysis
+- **synthesis/priority-recommendations.md**: Prioritized action items
+
+**Document Structure**:
+```
+.workflow/WFS-[topic]/.brainstorming/
+├── auto-session.json           # Session metadata and agent tracking
+├── system-architect-context.md # User responses for system-architect
+├── system-architect-template.md# Loaded role template
+├── system-architect/           # Dedicated agent outputs
+│   ├── analysis.md
+│   ├── recommendations.md
+│   └── deliverables/
+├── ui-designer-context.md      # User responses for ui-designer
+├── ui-designer/                # Dedicated agent outputs
+│   └── analysis.md
+└── synthesis/                  # Multi-role integration
+    ├── integrated-analysis.md
+    ├── consensus-matrix.md
+    └── priority-recommendations.md
+```
+
+## Reference Information
+
+### Role Processing Schema (Sequential Architecture)
+Each role processing follows dedicated agent pattern:
+- **role**: Selected planning role name
+- **template**: Loaded from cli-templates/planning-roles/
+- **context**: User responses with validation
+- **agent**: Dedicated conceptual-planning-agent instance
+- **output**: Role-specific analysis directory
+
+### File Structure Reference
+**Architecture**: @~/.claude/workflows/workflow-architecture.md
+**Role Templates**: @~/.claude/workflows/cli-templates/planning-roles/
+
+### Execution Integration
+Documents created for synthesis and action planning:
+- **auto-session.json**: Agent tracking and session metadata
+- **[role]-context.md**: Context loading for role analysis
+- **[role]/analysis.md**: Role-specific analysis outputs
+- **synthesis/**: Multi-role integration for comprehensive planning
+
+## Error Handling
+- **Role selection failure**: Default to `product-manager` with explanation
+- **Context validation failure**: Re-prompt with minimum requirements
+- **Agent execution failure**: Role-specific retry with corrected context
+- **Template loading issues**: Graceful degradation with fallback questions
+- **Multi-role conflicts**: Synthesis agent handles disagreement resolution
+
+## Quality Standards
+
+### Dedicated Agent Excellence
+- **Single role focus**: Each agent handles exactly one role - no multi-role assignments
+- **Complete context**: Each agent receives comprehensive role-specific context
+- **Sequential processing**: Roles processed one at a time with full validation
+- **Dedicated output**: Each agent produces role-specific analysis and deliverables
+
+### Context Collection Excellence
+- **Role-specific questioning**: Targeted questions for each role's domain expertise
+- **Context validation**: Verification before agent submission to ensure completeness
+- **User guidance**: Clear explanations of role perspective and question importance
+- **Response quality**: Minimum response requirements with re-prompting for insufficient detail