brainstorming

The Brainstorming skill transforms vague concepts into structured technical designs through structured, iterative inquiry. Rather than generating a massive document immediately, it focuses on precise discovery, gathering context from the existing codebase before proposing distinct architectural approaches. It operates through a conversational feedback loop, presenting design segments for validation to ensure alignment before finalizing the architecture. By adhering to YAGNI principles and requiring explicit confirmation at each stage, the skill minimizes wasted effort and prevents scope creep. The final output is a formal specification document saved to the project repository, providing a clear roadmap for implementation. This approach ensures that every proposed component, data flow, and testing requirement is intentionally vetted against the project's specific constraints and objectives.

When to Use This Skill

• •Defining the architecture for a new feature module
• •Refactoring an existing component to address performance issues
• •Evaluating different database schemas for a data-heavy application
• •Planning the integration of a third-party API or service

How to Invoke This Skill

Example prompts that trigger this skill in Claude Code, Cursor, or Antigravity:

• “Let's brainstorm a design for a new user authentication system
• “How should we architect the search functionality for this repo?
• “Help me plan the data flow for the incoming webhook processor
• “I have an idea for a new module, let's work through the design
• “Brainstorm approaches for upgrading our current logging service

Pro Tips

• 💡**Be specific with initial prompts:** The clearer your starting idea, the more focused and relevant the agent's questions and design proposals will be.
• 💡**Engage iteratively:** Don't rush the process. Respond thoughtfully to each question and provide feedback on design sections to guide the agent effectively.
• 💡**Use it for discovery, not just validation:** Leverage the agent's ability to propose multiple approaches and trade-offs to uncover solutions you might not have considered.

What this skill does

• •Context-aware project analysis using existing files and repository history
• •Single-question refinement loops to clarify project goals and constraints
• •Comparative design assessment offering multiple technical approaches
• •Incremental validation of design segments to catch errors early
• •Automated creation of timestamped design documentation files

When not to use it

• ✕Performing simple bug fixes or minor code cleanup
• ✕Making high-level business strategy decisions outside of code architecture
• ✕Writing code implementation without a predefined design plan

Example workflow

1. Scan current project files and commit history for context
2. Ask clarifying questions about project goals and success criteria
3. Propose three distinct technical approaches with trade-offs
4. Draft the design in 300-word sections with pause-and-validate prompts
5. Write the finalized specification to docs/plans/ and commit to git

Prerequisites

• –Existing codebase or repository
• –Basic project requirements or rough idea

Pitfalls & limitations

• !Over-engineering if YAGNI principles are ignored
• !Getting stuck in iterative loops if questions are too vague
• !Lack of technical clarity if the initial context search is incomplete

FAQ

How it compares

Unlike a generic prompt that might output an entire design at once, this skill enforces a step-by-step validation process that prevents hallucinated or misaligned architectural choices.

Source & trust

Recommended Rules

Recommended Workflows

Recommended MCP Servers