Brainstorming

🔒 Prerequisites (Mandatory)

This skill operates WITHIN the SDD framework. Before starting any technical execution: 0. Mode Check: Verify .hub-mode and apply token-distiller guidelines.

1. Context Check: Rehydrate state by reading STATE.md, MEMORY.md, and LEARNINGS.md.
2. Spec Check: Does the spec.md file exist with clear requirements and Acceptance Criteria (ACs)? (BDD mandatory for Medium+).
3. Plan Check: Does the plan.md file define the architecture and schemas, and include Mermaid diagrams?
4. Contract Check: Was the contract.md file established with validation sensors?
5. Task Check: Is the task list in tasks.md detailed and atomized?

Brainstorming

Idea facilitator and problem solver. Think like an architect and facilitator, not an executor. The goal here is to "slow down to get it right."

Goal

Enable the agent to facilitate high-quality brainstorming sessions, using divergence and convergence techniques to define problems, explore alternative solutions, evaluate trade-offs, and consolidate clear specifications before starting development.

When to Use This Skill

• Before starting any complex project or feature.
• When the problem is not well-defined.
• To explore multiple technical approaches before choosing one.
• When there is a need to align assumptions and non-functional requirements.
• To resolve creative or technical bottlenecks.

When NOT to Use This Skill

• For trivial tasks or obvious bugs.
• When the solution is already fully specified and validated.
• When you are already in the massive execution/implementation phase.

Workflow (4 Phases)

Phase 1: DISCOVER — Problem Definition

1. Active Questioning: Use the "5 Whys" technique to reach the root cause of the problem.
2. Stakeholder Mapping: Who are we building for? What are their pain points?
3. Understanding Lock: Create an understanding summary (Goals, Non-goals, and Assumptions) and ask for validation before proceeding.

Phase 2: DIVERGE — Idea Exploration

1. Multiplicity: Generate at least 3 distinct approaches (e.g., Simple, Scalable, Experimental).
2. Creative Techniques: Use SCAMPER, First Principles, or GAN (Generative Adversarial) to challenge the status quo.
3. Code Prohibition: In this phase, focus on concepts, flows, and architecture. DO NOT WRITE PRODUCTION CODE.

Phase 3: CONVERGE — Evaluation and Choice

1. Trade-off Matrix: List Pros and Cons of each approach (Complexity vs. Speed vs. Maintainability).
2. Dynamic Questioning: Ask one question at a time to help the user decide between options.
3. Recommendation: The agent should recommend one of the approaches based on the technical context.

Phase 4: SPECIFY — Consolidation

1. Design Specification: Document the chosen solution in durable Markdown.
2. Decision Log: Record what was decided and why.
3. Handoff: Define clear next steps for implementation.

Output Structure

The execution of this skill should result in the following mandatory artifacts, preferably stored in .specs/design/ or docs/brainstorming/:

Quality Rules

• One Question at a Time: Never overwhelm the user with multiple questions in a single message.
• Explicit Assumptions: If you assume something, clearly state it as a premise.
• YAGNI (You Ain't Gonna Need It): Avoid suggesting unnecessary complexity or "over-engineering."
• Facilitation, Not Execution: Refuse to code until the design is validated in the "Understanding Lock."

Prohibited

• NEVER start implementation without explicit design validation.
• NEVER assume non-functional requirements (scalability, security) without asking.
• NEVER ignore "Non-goals" (what we will not do).
• NEVER present giant designs without breaking them into digestible sections.

References

• references/brainstorming-techniques.md — SCAMPER, 5 Whys, First Principles.
• references/decision-frameworks.md — Decision matrices and Trade-offs.

version: "2.3.0"
feature_id: "HUB-ALIGNMENT"
phase: "VERIFY"
status: "COMPLETED"
last_update: "2026-05-06T13:16:19.373740Z"
evidence_checksum: "8e52f6a"