What problem does it solve? Discovery material accumulates faster than anyone reads it, so the same patterns get rediscovered repeatedly while the strongest cross-source insights are never found. This Skill turns a pile of chats, feedback rounds, or research packs into a short, ranked, sourced list of insights deduplicated against an existing ledger. ## Core Features & Use Cases - Scored ranking: Scores every candidate on convergence, emergence, surprise, tension, and decision-relevance, keeping only the top three to seven. - Ledger deduplication: Checks candidates against the existing insights ledger, marking re-confirmations and strengthening existing entries instead of duplicating them. - Evidence-capped confidence: Assigns forming, firming, or firm confidence based on who and how many sources back each claim, with founder-only evidence capped at forming. - Use Case: Point it at a folder of user interview notes and ask what stands out; it returns a ranked list of insights with evidence, confidence levels, and the decision each one would change. ## Quick Start Ask the agent to mine the attached research notes and pull out the insights that would change a product decision.
