trellis-brainstorm

To structure AI brainstorming for clear task scope, guide the AI through collaborative requirements discovery to seed a task directory and a PRD, ensuring initial ideas are captured immediately before any coding begins.

The best way to create a PRD and define MVP scope with unclear requirements is to use a diverge-then-converge workflow that proactively explores edge cases before narrowing down to a structured PRD with goals, assumptions, and acceptance criteria.

To document assumptions and acceptance criteria for a new feature, trigger a brainstorming workflow that automatically scaffolds a task directory and a PRD, capturing goals, questions, and technical choices alongside the requirements.

Yes, AI can automatically research technical choices and derive answers from existing repo code and conventions before asking questions, ensuring it presents concrete options and trade-offs rather than requesting basic information.

You should use a requirements discovery workflow instead of directly coding when a new feature has multiple viable approaches, unclear requirements, or complex task descriptions that need deliberate exploration and MVP scoping.

This brainstorming approach captures both immediate MVP scope and future expansion plans by proactively considering future evolution and related scenarios during the diverge phase, then converging to a structured PRD and MVP plan.