prd-generator

To generate a PRD from an existing codebase, use a reverse-engineering mode that scans repository artifacts to draft an 11-section product requirements document. The output focuses on product behavior and includes confidence tags with explicit review markers for inferred content.

The best way to validate a PRD for completeness is to score each section against a standardized rubric, producing a graded validation report. This process identifies gaps in problem definition, users, features, constraints, and MVP scope, providing targeted recommendations for improvement.

Yes, you can create a complete product requirements document by using an interactive builder that collects project details and maps them to a consistent 11-section schema. This ensures your PRD captures problems, users, features, constraints, MVP scope, and open questions.

Reverse-engineering prevents hallucinated features by enforcing anti-hallucination constraints that trace inferred features directly to code evidence. It adds explicit confidence tags and review markers to clearly distinguish verified product requirements from unverified claims.

You need to reverse-engineer product requirements when a team lacks documentation but needs a clear product plan based on existing implementation. This process drafts what the product should do by scanning repository artifacts rather than focusing on how to implement it.

Yes, PRD generation works for grading an incomplete product plan by running a validator that scores existing sections against a completeness rubric. It outputs a graded validation report highlighting missing sections and offering targeted recommendations to fill gaps.