pdf-remix — extract a PDF's mold, then remix it with new content

reel-discovery finds a winning video and lets you restyle it. pdf-remix does the same for a PDF: it deconstructs a premium document into a reusable mold (design system + per-page archetypes + narrative arc) and then rebuilds a new document — new topic, new copy, freshly generated imagery — that follows the mold faithfully. Text stays crisp and editable (HTML → Chromium PDF) while asset-generator produces the visuals.

Two phases, one skill:

ANALYZE   reference.pdf ─▶ blueprint.json + blueprint.md + content_template.json
REMIX     content.json (+brand +refs) ─▶ assets/*.png ─▶ output.pdf

The skill is brand-agnostic: any reference PDF, any topic, any brand. Point it at reference photos (e.g. a person) and it will generate that person into poses that fit the new pages (a cover portrait, etc.).