What problem does it solve? Writing PRDs often suffers from two failure modes: AI assistants inventing plausible-sounding content the user never said, and hypotheses being stated without any numerical evidence. This Skill enforces a faithful-documentation principle (only write what the user said, mark everything else as [TBD]) and adds a data collection and analysis phase so every claim in the PRD is backed by verifiable numbers. ## Core Features & Use Cases - Faithful PRD Skeleton Creation: Classifies user statements into explicit facts, implied directions, and unmentioned items, then builds a PRD draft with [TBD] placeholders instead of guesses. - Multi-Source Data Collection: Pulls evidence from email delivery tool history (send dates, notification counts, CTR), BI tools via SQL (cohort analysis, overall trends), GitHub PRs/Issues, and internal documentation. - Claim-Data Reconciliation: Compares user hypotheses against actual data and reports discrepancies in conversation rather than silently rewriting the PRD. - Use Case: A product manager wants to write a PRD showing the effect of three feature releases. The Skill drafts the PRD from their verbal explanation, generates SQL queries for cohort analysis, cross-references email campaign send dates, and fills in week-over-week increments with full source traceability. ## Quick Start Ask the assistant to create a data-backed PRD for your feature initiative, then describe the initiative verbally and let it build the skeleton and identify which data sources to query.
