bmad-create-prd

Creating a PRD from scratch is handled by a step-file workflow that sequentially guides you through vision, success criteria, user journeys, requirements, and scope, ultimately appending approved content to a markdown artifact.

A complete PRD should include vision, success criteria, user journeys, domain constraints, scope, and functional requirements, serving as a dense, readable capability contract for both human teams and downstream LLM consumption.

Yes, the workflow is brownfield-aware, automatically detecting existing project documentation and adapting the PRD generation steps to focus specifically on changes and additions rather than starting from zero.

Yes, Python3 is required as a dependency to execute the workflow, specifically running the resolve_customization.py script to resolve customizations and load persistent facts before step execution begins.

The best way to define scope and non-functional requirements is through a sequential step execution process that enforces structured discovery, preventing premature content generation until all requirement constraints are properly collected and approved.

Sequential step execution is enforced to prevent skipping steps or generating content prematurely, ensuring that persistent facts and customization rules are loaded before any vision, scope, or requirement content is appended to the final document.