Brainstorming

Domain skill for multi-option exploration and collaborative decision-making before implementation.

Iron Law

EXPLORE BEFORE COMMITTING. The cost of choosing wrong is 10x the cost of exploring options.

Every hour spent exploring saves a day of rework from the wrong approach.

When to Brainstorm

• Multiple valid approaches exist (and you know it)
• Requirements are unclear or ambiguous
• New territory — no existing patterns to follow
• User says "help me think through" or "what are our options"
• You catch yourself defaulting to the first idea

Exploration Process

Track the brainstorming lifecycle with tasks:

TaskCreate("Clarify goal", "Confirm what, why, who before generating options")
TaskCreate("Check existing patterns", "Search codebase for how similar problems are solved")
TaskCreate("Generate approaches", "Produce 2-4 distinct approaches with pros, cons, effort, risk")
TaskCreate("Compare and recommend", "Build comparison table, highlight trade-offs, recommend")
TaskCreate("User decision", "Present recommendation, get user choice, log decision")

1. Clarify the Goal

TaskUpdate("Clarify goal", status: "in_progress")

Before generating options, confirm you understand:

• What needs to happen (outcome, not implementation)
• Why it needs to happen (motivation, constraints)
• Who is affected (users, systems, team)

Use AskUserQuestion if the goal is ambiguous. Don't guess.

TaskUpdate("Clarify goal", status: "completed")

2. Check Existing Patterns

TaskUpdate("Check existing patterns", status: "in_progress")

Before proposing new approaches, see what the codebase already does:

# How does the project handle similar problems?
grep -r "relevant_pattern" --include="*.{ts,js,py,rb}" -l
git log --oneline --all --grep="related keyword" | head -10

Existing patterns get priority — consistency has value.

TaskUpdate("Check existing patterns", status: "completed")

3. Generate 2-4 Approaches

TaskUpdate("Generate approaches", status: "in_progress")

For each approach, fill out:

Rules:

• Minimum 2 approaches, maximum 4
• Approaches must be genuinely distinct (not minor variations)
• Include the "do nothing" or "simplest possible" option when applicable
• Each approach must be implementable — no hand-waving

TaskUpdate("Generate approaches", status: "completed")

4. Compare Trade-offs

TaskUpdate("Compare and recommend", status: "in_progress")

Build a comparison table across key dimensions:

TaskUpdate("Compare and recommend", status: "completed")

5. Recommend and Decide

TaskUpdate("User decision", status: "in_progress")

• State your recommendation with rationale
• Highlight the most important trade-off
• Let the user choose — recommendation is not a decision
• Use AskUserQuestion for the final decision

TaskUpdate("User decision", status: "completed") after user selects an approach. Use TaskList to confirm the full brainstorming lifecycle completed.

Trade-Off Dimensions

Common dimensions to evaluate (pick the relevant ones):

Anti-Patterns

Rationalization Prevention