Walks a PM through the layer-1 ritual of reading raw transcripts of an AI feature in action — what Anthropic calls "manual testing, dogfooding, and intuition" — before formalizing a structured eval. Surfaces failure patterns, drafts task candidates (including balanced negative cases per Anthropic St
Walks a PM through the layer-1 ritual of reading raw transcripts of an AI feature in action — what Anthropic calls "manual testing, dogfooding, and intuition" — before formalizing a structured eval. Surfaces failure patterns, drafts task candidates (including balanced negative cases per Anthropic Step 3), and produces a verdict on whether the feature is ready for /eval. Use when the user says "I w
Cette entrée ne publie aucun paquet npm : Forge n'a donc pas d'arbre de dépendances pour elle. C'est une lacune de couverture — pas une affirmation qu'elle n'a aucune dépendance.