What problem does it solve? Product bets often fail the "how do you know?" challenge because evidence is aggregated into unsourced claims, quotes get paraphrased, and counter-evidence is quietly dropped. This Skill provides the rules for building Evidence sections on initiatives, PRDs, and intake investigations that preserve source attribution, verbatim quotes, and honest counter-evidence. ## Core Features & Use Cases - Evidence pyramid ranking: Classifies signals by trust tier, from verbatim customer quotes down to internal hunches, and requires each Evidence section to name its tier. - Attribution-preserving aggregation: Rolls up evidence across many linked intakes by summarizing counts while keeping per-source links and verbatim quotes intact. - Synthesize-then-compare discipline: Structures agent synthesis as a proposed reading that the PM reconciles against their own independent read, with theme-to-verbatim links preserved. - Use Case: When promoting 12 intake tickets about SAML requests into a single initiative, use this Skill to generate an Evidence section that shows the aggregate count, lists each source intake with its verbatim quote, separates interpretation into the Bet section, and includes a counter-evidence subsection naming the segment where the bet does not hold. ## Quick Start Ask the agent to synthesize the linked intakes for an initiative into an Evidence section with verbatim quotes, source attribution, and a counter-evidence subsection.
