brainstorming_skill

Overview

This skill refines rough ideas into fully formed designs through structured Socratic questioning, alternative exploration, and incremental validation. It prioritizes understanding constraints, proposing multiple approaches, and presenting the design in bite-sized sections for user validation. I announce at the start that I am using the brainstorming skill to refine your idea into a design.

How this skill works

Begin by creating a task tracker and dispatching investigators to inspect the codebase or external sources when needed. Phase 1 gathers purpose, constraints, and success criteria by asking one clear question at a time. Phase 2 proposes 2–3 researched approaches with trade-offs and asks you to choose. Phase 3 presents the design in 200–300 word sections covering architecture, components, data flow, and contracts, then asks for validation before proceeding.

When to use it

• Before writing code or implementation plans to clarify goals and constraints
• When you need structured exploration of 2–3 architectural approaches
• To validate boundaries, interfaces, and component contracts without implementation details
• When a project has unknowns that require targeted research or codebase inspection
• If you want incremental feedback to avoid late-stage rework

Best practices

• Create a task tracker at the start and mark phases in_progress/completed
• Ask only one useful question at a time and prefer AskUserQuestion for clear choices
• Delegate codebase inspection and external research to specialized agents instead of deep research yourself
• Always present at least two alternative approaches and surface trade-offs
• Present designs in short sections and ask open-ended validation questions after each

Example use cases

• Designing a new authentication flow while checking existing codebase patterns
• Choosing between event-driven, synchronous, or hybrid architectures for a feature
• Refining product requirements into clear component contracts before implementation
• Exploring library options and trade-offs when no project pattern exists
• Validating a proposed API surface and data flow with stakeholders iteratively