project-brainstorming（项目头脑风暴）

When To Use

• Starting a new project without clear requirements
• Exploring problem space before specification
• Need to compare multiple approaches systematically
• Validating project feasibility and scope
• Documenting decision rationale for stakeholders
• Need to clarify the core problem being solved

When NOT To Use

• Requirements and specification already exist (use Skill(attune:project-planning) instead)
• Refining existing specs (use Skill(attune:project-specification) instead)
• Project scope is well-defined (jump to /attune:project-init)
• Mid-project pivots (use Skill(attune:war-room) for strategic decisions)

Integration

With superpowers:

• Delegates to Skill(superpowers:brainstorming) for Socratic method
• Augments with project-specific patterns
• Uses project brainstorm templates

Without superpowers:

• Standalone questioning framework
• Project-focused ideation patterns
• Structured output templates

War Room Integration (REQUIRED):

• After Phase 3 (Approach Generation), automatically invokes Skill(attune:war-room)
• All brainstorming context passed to War Room for expert deliberation
• War Room provides multi-LLM pressure testing and synthesis
• Only bypassed for Type 2 decisions (RS ≤ 0.40) with explicit user confirmation

Brainstorming Framework

Phase 1: Problem Definition

Socratic Questions:

1. What problem are you solving?
2. Who experiences this problem?
3. What makes this problem worth solving now?
4. What happens if this problem isn't solved?
5. What existing solutions have been tried?

Output: Problem statement in docs/project-brief.md

Template:

## Problem Statement

**Who**: [Target users/stakeholders]
**What**: [The problem they face]
**Where**: [Context where problem occurs]
**When**: [Frequency/timing of problem]
**Why**: [Impact of the problem]
**Current State**: [Existing solutions and limitations]

Verification: Run the command with --help flag to verify availability.

Phase 2: Constraint Discovery

Questions:

1. What are non-negotiable technical constraints?
2. What are resource constraints (time, budget, team)?
3. What integration points are required?
4. What compliance/regulatory requirements apply?
5. What are success criteria and failure modes?

Output: Constraints matrix

Template:

## Constraints

### Technical
- [Constraint 1 with rationale]
- [Constraint 2 with rationale]

### Resources
- **Timeline**: [Duration with milestones]
- **Team**: [Size and skills]
- **Budget**: [If applicable]

### Integration
- [Required system 1]
- [Required system 2]

### Compliance
- [Requirement 1]
- [Requirement 2]

### Success Criteria
- [ ] [Measurable criterion 1]
- [ ] [Measurable criterion 2]

Verification: Run the command with --help flag to verify availability.

Phase 3: Approach Generation

Technique: Generate 3-5 distinct approaches

For each approach:

• Clear description (1-2 sentences)
• Technology stack
• Pros (3-5 points)
• Cons (3-5 points)
• Risks (2-3 points)
• Estimated effort
• Trade-offs

Template:

## Approach [N]: [Name]

**Description**: [Clear 1-2 sentence description]

**Stack**: [Technologies and tools]

**Pros**:
- [Advantage 1]
- [Advantage 2]
- [Advantage 3]

**Cons**:
- [Disadvantage 1]
- [Disadvantage 2]
- [Disadvantage 3]

**Risks**:
- [Risk 1 with likelihood]
- [Risk 2 with likelihood]

**Effort**: [S/M/L/XL or time estimate]

**Trade-offs**:
- [Trade-off 1 with mitigation]
- [Trade-off 2 with mitigation]

Verification: Run the command with --help flag to verify availability.

Design for Isolation:

When generating approaches, evaluate each against two isolation tests:

1. Comprehension test: Can someone understand what each unit does without reading its internals? If a unit requires reading implementation details to understand its purpose, the boundary is wrong.
2. Change test: Can you change a unit's internals without breaking its consumers? If changing implementation details forces changes elsewhere, the interface is leaking.

File size as design signal: Files exceeding 500 lines (Python/Go) or 300 lines (JavaScript/TypeScript) often indicate a unit is doing too much. This is a design smell, not just a style issue. When flagging large files, suggest extracting specific concerns (e.g., "Extract validation logic into a separate module to improve testability").

Verification: Run the command with --help flag to verify availability.

Phase 3.5: War Room Deliberation (REQUIRED)

Automatic Trigger: After generating approaches, MUST invoke Skill(attune:war-room) for expert deliberation

When War Room is invoked:

• All brainstorming context (problem, constraints, approaches) automatically passed to War Room
• Expert panel reviews, challenges, and pressures each approach
• Reversibility assessment conducted
• Multi-LLM deliberation identifies blind spots
• Supreme Commander provides synthesis with rationale

Command:

# Automatically invoked from brainstorm - DO NOT SKIP
/attune:war-room --from-brainstorm

War Room Output:

• Reversibility Score (RS) and decision type
• Red Team challenges for each approach
• Premortem analysis on selected approach
• Supreme Commander Decision document
• Implementation orders and watch points

Bypass Conditions (ONLY skip war room if ALL true):

• RS ≤ 0.40 (Type 2 decision - clearly reversible)
• Single obvious approach with no meaningful trade-offs
• Low complexity with well-documented pattern
• User explicitly declines after being shown RS assessment

Proceed to Phase 4 only after War Room completes

Phase 4: Approach Comparison

Comparison Matrix:

Scoring: 🟢 = Good, 🟡 = Acceptable, 🔴 = Concern

Phase 5: Decision & Rationale

Selection Criteria:

1. Alignment with constraints (must satisfy all)
2. Risk vs. reward balance
3. Team capability and experience
4. Time to value
5. Long-term maintainability

Template:

## Selected Approach: [Approach Name] ⭐

### Rationale
[2-3 paragraphs explaining why this approach was selected]

Key decision factors:
- [Factor 1]
- [Factor 2]
- [Factor 3]

### Trade-offs Accepted
- **Trade-off 1**: [Description] → Mitigation: [Strategy]
- **Trade-off 2**: [Description] → Mitigation: [Strategy]

### Rejected Approaches
- **Approach X**: Rejected because [reason]
- **Approach Y**: Rejected because [reason]

Verification: Run the command with --help flag to verify availability.

Output: Project Brief

Final output saved to docs/project-brief.md:

# [Project Name] - Project Brief

**Date**: [YYYY-MM-DD]
**Author**: [Name]
**Status**: Draft | Approved

## Problem Statement
[From Phase 1]

## Goals
1. [Primary goal]
2. [Secondary goal]
3. [Tertiary goal]

## Constraints
[From Phase 2]

## Approach Comparison
[From Phase 3 & 4]

## War Room Decision
[From Phase 3.5 - includes RS assessment, Red Team challenges, premortem]

## Selected Approach
[From Phase 5, informed by War Room synthesis]

## Next Steps
1. `/attune:specify` - Create detailed specification
2. `/attune:blueprint` - Plan architecture and tasks
3. `/attune:project-init` - Initialize project structure

Verification: Run the command with --help flag to verify availability.

Questioning Patterns

Socratic Method

Clarification:

• "What do you mean by [term]?"
• "Can you give an example?"
• "What is the difference between X and Y?"

Probing Assumptions:

• "What are you assuming about [aspect]?"
• "Why do you think that assumption is valid?"
• "What if that assumption is wrong?"

Probing Reasoning:

• "Why do you think this approach is best?"
• "What evidence supports this?"
• "Are there alternative explanations?"

Questioning Viewpoints:

• "What would [stakeholder] think about this?"
• "What are the counterarguments?"
• "How might this fail?"

Probing Implications:

• "What happens if we choose this approach?"
• "What are the long-term consequences?"
• "What does this commit us to?"

Constraint-Based Thinking

Must Have (Non-negotiable):

• What absolutely must be true for success?
• What constraints cannot be changed?

Should Have (Important):

• What would significantly increase success?
• What preferences matter most?

Could Have (Nice to have):

• What would be beneficial but not critical?
• What can we defer or drop if needed?

Won't Have (Explicit exclusions):

• What are we explicitly NOT doing?
• What scope boundaries prevent creep?

Red Flags to Surface

During brainstorming, watch for:

• ⚠️ Vague problem statements ("make it better")
• ⚠️ Unclear success criteria
• ⚠️ Hidden assumptions about users or technology
• ⚠️ Single approach bias (not exploring alternatives)
• ⚠️ Scope creep in requirements
• ⚠️ Unrealistic constraints or timelines
• ⚠️ Missing stakeholder perspectives

Session State Management

Save session to .attune/brainstorm-session.json:

{
  "session_id": "20260102-143022",
  "started_at": "2026-01-02T14:30:22Z",
  "current_phase": "approach-selection",
  "problem": {
    "statement": "...",
    "stakeholders": ["..."]
  },
  "constraints": {
    "technical": ["..."],
    "resources": {"timeline": "...", "team": "..."}
  },
  "approaches": [
    {
      "name": "...",
      "pros": ["..."],
      "cons": ["..."]
    }
  ],
  "selected_approach": null,
  "decisions": {}
}

Verification: Run the command with --help flag to verify availability.

Phase 6: Workflow Continuation (REQUIRED)

Automatic Trigger: After Phase 5 (Decision & Rationale) completes and docs/project-brief.md is saved, MUST auto-invoke the next phase.

When continuation is invoked:

1. Verify docs/project-brief.md exists and is non-empty
2. Display checkpoint message to user: Brainstorming complete. Project brief saved to docs/project-brief.md. Proceeding to specification phase...
3. Invoke next phase: Skill(attune:project-specification)

Bypass Conditions (ONLY skip continuation if ANY true):

• --standalone flag was provided by the user
• docs/project-brief.md does not exist or is empty (phase failed)
• User explicitly requests to stop after brainstorming

Do NOT prompt the user for confirmation — this is a lightweight checkpoint, not an interactive gate. The user can always interrupt if needed.

Phase 6.5: Spec Review Gate

Automatic Trigger: After Phase 6 saves the project brief, and before invoking the next phase, run the spec review loop.

Procedure:

1. Load modules/spec-review-loop.md for the review prompt template
2. Dispatch haiku-model subagent with the spec content
3. If ISSUES FOUND: fix issues, re-dispatch (max 3 iterations)
4. If APPROVED or 3 iterations exhausted: proceed to Phase 6 continuation

Bypass Conditions:

• --standalone flag was provided
• --skip-review flag was provided
• Spec document is under 200 words (too small to review)

Related Skills

• Skill(superpowers:brainstorming) - Socratic method (if available)
• Skill(attune:war-room) - REQUIRED AUTOMATIC INTEGRATION - Invoked after Phase 3 for multi-LLM deliberation
• Skill(imbue:scope-guard) - Scope creep prevention
• Skill(attune:project-specification) - AUTO-INVOKED next phase after brainstorming
• Skill(attune:mission-orchestrator) - Full lifecycle orchestration

Related Commands

• /attune:brainstorm - Invoke this skill
• /attune:specify - Next step in workflow
• /imbue:feature-review - Worthiness assessment

Examples

See /attune:brainstorm command documentation for complete examples.