cds-brainstorming

name	cds-brainstorming
description	Clarify what should be built before planning begins. Use when the problem is under-defined, the intended outcome is unclear, or there are materially different directions to explore before committing to a plan.

Brainstorming

Purpose

Clarify what should be built before planning begins. Resolve ambiguity in the problem, the intended outcome, or the approach. Land on a clear direction that planning can execute against.

When To Use

• The problem itself is still under-defined.
• The intended outcome is unclear or has multiple valid interpretations.
• There are materially different product directions to explore before committing to a plan.
• Option comparison is needed to narrow down the right approach in an unfamiliar problem space.

When Not To Use

• Requirements are specific and well-scoped. Go to planning.
• The task is a known fix or implementation with no meaningful design choice.
• The problem is clear but needs implementation option comparison. Go to planning.

Core Principles

1. Clarify before committing. The goal is to resolve ambiguity in the problem, the outcome, or the direction. Do not proceed to planning until the problem is defined well enough to plan against.
2. Land on a decision. Every brainstorm must end with a clear direction. Exploration without a conclusion is wasted work.
3. Stay focused on what, not how. Implementation details belong in planning. Brainstorming defines the problem and the chosen direction.
4. Match the problem's complexity, not more. Capture the real difficulty of the problem faithfully. Do not add speculative requirements, premature abstractions, or hypothetical future concerns that the problem itself does not demand.
5. One question at a time. Do not overwhelm with multiple questions. Confirm understanding before moving forward.

Workflow

1. Assess clarity. Is the problem defined well enough to plan? If yes, skip to planning. If not, proceed.
2. Research the problem space. Read relevant code, docs, specs, and existing patterns. Understand what exists and what the problem touches.
3. Interview for clarity. Ask focused questions one at a time across the prescribed categories. Confirm understanding before moving forward.
4. Explore options. When multiple directions are viable, compare them. Include only real options that meet the quality bar. Land on a recommendation.
5. Capture the decision. Write the brainstorm document with uniform headings.

Question Categories

These categories should be covered during the interview step. Not every category applies to every brainstorm — mark N/A when not relevant.

1. Problem definition. What is the actual problem and why does it matter now?
2. User intent. Who needs this and what outcome do they expect?
3. Success criteria. What does success look like concretely?
4. Constraints. Known limits — technical, timeline, scope, dependencies.
5. Existing patterns. What already exists that is relevant? What has been tried before?
6. Edge cases and risks. What could go wrong? What should explicitly not happen?

Output Location

Brainstorm documents are written to docs/brainstorms/ with date-prefixed filenames:

docs/brainstorms/YYYY-MM-DD-<descriptive-name>.md

Brainstorm Template

---
title: <brainstorm title>
date: YYYY-MM-DD
status: exploring | decided
---

# <Brainstorm Title>

## Problem Statement

What problem is being solved and why it matters.

## User Intent

Who needs this and what outcome they expect.

## Success Criteria

How to know the solution is working.

## Constraints

Known limits that shape the direction.

## Options Explored

Viable directions considered. What each offers and its key tradeoff.

### Option A: <name>

<description and tradeoff>

### Option B: <name>

<description and tradeoff>

### Selected: <option>

<why this direction>

## Architecture Sketch

Rough sketch of boundaries and data flow for the selected direction. Use Mermaid fenced code blocks.

```mermaid
graph LR
    A[Service] --> B[Service]

Key Decisions

Decisions made during exploration with rationale.

Open Questions

Unresolved questions that planning or implementation must answer.

Next Steps

What happens next — typically planning.

## Constraints On This Skill

- This skill explores and documents decisions only. It must never generate implementation code.
- Questions are asked one at a time. Confirm understanding before moving forward.
- All template headings are mandatory. Mark sections N/A when not relevant.
- File references use paths relative to the repository root.
- Keep instructions agent-agnostic. No tool-specific or platform-specific references.