What problem does it solve? After producing a weekly batch of Instagram Reels, insights about which hooks retained viewers, which archetypes converted, and which quality gates were violated get lost between batches. This Skill closes the feedback loop by generating a structured retrospective document that feeds directly into the next batch's planning phase. ## Core Features & Use Cases - Batch inventory and metrics collection: Reads the usage registry and approved output files to map each Reel's Trojan-horse family, archetype, belief, headline, and CTA mode, then records manually pasted Instagram metrics (retention, shares, saves, non-follower reach) with explicit pending placeholders instead of invented numbers. - Pattern and risk analysis: Cross-references violated quality gates against the rules framework, measures family/archetype diversity, and compares conversion-mode versus engagement-mode Reels to detect reach erosion outside the follower bubble. - Writing lesson proposals: Compares V1 drafts against final approved scripts, extracts before/after edit pairs, and proposes new writing lessons for the knowledge base, which are only saved after explicit user approval. - Use Case: At the end of a 5-Reel week, run the retrospective to produce trabalho/retrospectivas/AAAA_SEMNN.md with metrics, diversity coverage, lessons learned, and concrete keep/change/add/remove recommendations for the next batch. ## Quick Start Ask the assistant to close the retrospective for this week's Reel batch, pasting any Instagram metrics you already have from the app.
