dvcrn Brainstorming — Design Before Code
When to Use
Use BEFORE starting any non-trivial implementation. If you're about to write code for a new feature, architecture change, or multi-file modification, STOP and brainstorm first.
Hard Gate
Do NOT write implementation code until you have:
- Explored the codebase to understand current state
- Proposed 2-3 approaches with trade-offs
- Received approval on the chosen approach
Process
Step 1: Explore Context
Before proposing anything:
- Read recent commits to understand recent direction
- Search the codebase for related code and patterns
- Check existing documentation for prior decisions
- Understand what exists before proposing changes
Step 2: Propose Approaches
Present 2-3 approaches with trade-offs and your recommendation:
DESIGN: [Feature Name]
Approach 1 (Recommended): [description]
- Pro: [benefit]
- Con: [cost]
Approach 2: [description]
- Pro: [benefit]
- Con: [cost]
Approach 3: [description]
- Pro: [benefit]
- Con: [cost]
My recommendation: Approach 1 because [reason].
Waiting for approval before proceeding.
Step 3: Wait for Approval
Do NOT proceed until the decision-maker responds. If no response within the current session, note the pending design and move on to other work.
Step 4: Transition to Planning
Once approved, create a detailed implementation plan before writing code. Use craftwork:writing-plans to structure the plan.
Anti-Patterns
- "This is too simple to need a design" — Simple tasks cause the most wasted work from unexamined assumptions.
- "I'll just figure it out as I go" — Implementation without direction leads to rework.
- "We already know how to do this" — Search for prior decisions first. Don't repeat past mistakes.
Key Principles
- YAGNI ruthlessly — remove unnecessary features from all designs
- Lead with your recommendation — don't present options without opinion
- Scale to complexity — a few sentences for simple tasks, detailed analysis for complex ones
- One question at a time — don't overwhelm with multiple questions