What problem does it solve? Understanding why viral Reels work is hard when you only watch them passively. This Skill turns a folder of reference Reels you provide into structured, actionable patterns by extracting audio, transcribing speech, and decomposing each video into a structural breakdown (hook, block-by-block skeleton, retention devices, bridge percentage, CTA, pacing). ## Core Features & Use Cases - Automated transcription pipeline: Extracts 16kHz mono audio with ffmpeg and transcribes with whisper, batching when many videos are present. - Structural decomposition: Produces a per-Reel ficha (hook, block skeleton, retention devices, bridge %, CTA, duration/rhythm), cross-referenced against your framework vocabulary files. - Pattern consolidation: Ranks recurring hook molds, structures, devices, and CTAs into a consolidated patterns document saved in trabalho/referencias/. - Use Case: You collected 10 high-performing Reels in your niche. Point the Skill at the folder and receive an index, transcripts, per-Reel structural sheets, and a consolidated patterns report to inform your next weekly batch. ## Quick Start Run /analisar-referencias with the path to your folder of .mp4 reference videos, for example by asking to analyze the Reels in your swipe folder.
