carousel

How to make a multi-style multi-slide social carousel (5-10 slides × N styles) where the deliverable is the baked-text image itself — no HyperFrames compose. A domain overlay on the standard pipeline: supplies the cover-first checkpoint (one cover per style → user review → fill the rest), the dual-ref cohesion rule (--ref mascot --ref {style}-01 on every fill slide), the mascot-fit rule (a clean cute mascot does NOT survive gritty/grunge styles — build a distressed character variant for those), the JSON prompt schema (reusable STYLE + QUALITY blocks per aesthetic, only scene + ui_elements varies per slide), and the gpt-5.4-image-2 serialize-with-transient-retry batch shape. Works for ANY mascot / brand + ANY set of aesthetics — zine / club / swiss / riso / punk / acid / vaporwave / clean DTC. USE WHEN the user asks for an Instagram / LinkedIn / TikTok carousel, "5-slide deck", "swipe-through", "multi-slide post", "story-style series", "N-style carousel". Multi-slide stills, not video. This is a niche SKILL (generalized), not a remix TEMPLATE. For "remix this exact carousel but swap X", use the remix path in docs/skills-vs-templates.md. For a single still, use /poster.