design-brief

Design Hand-off Brief

Write a precise, executable brief for the visuals supporting $ARGUMENTS, as a single artifact the person pastes straight into Claude Design (or any design tool). Output a spec another tool can follow — not the images.

Run this after the content is locked (Phase 3). If sections change later, this artifact is stale — regenerate it.

Principle: fewer graphics, the right ones

LinkedIn is the visual platform (the carousel is the content, shipped as a PDF). Substack is reading-and-email — image-heavy emails render poorly and hurt deliverability, so keep it lean. Medium rewards a strong hero plus a few section images. Spec what does a job; don't decorate.

Shared visual identity (state once, applies to every asset)

Open the artifact with one identity block so every image matches:

• Tone matches the content's framing. If the piece is evolutionary/positive, the visuals must be too — explicitly forbid warning aesthetics (red, cracks, alert icons) so nothing reads as ominous.
• Background — one consistent surface (e.g. warm neutral/cream).
• One accent color, reserved for the hero concept of the story; calm consistent neutrals for everything else.
• Card style — rounded cards, soft shadow; labels in monospace SMALL CAPS for a technical feel.
• Byline (bottom-right of each asset) — the author's name + tagline (carry the person's existing byline; ask if unknown).
• Variants — produce light (default) + dark versions of each image.
• Clean and legible — big type, generous whitespace, must survive being viewed tiny in a feed or email.
• Asset naming — number assets by section so they map to the post, e.g. light-assets/0-hero.png, light-assets/1-...png, with a matching dark-assets/ set.

Part A — Substack & Medium: one image per section

These two read as articles, so spec one image per major section, plus a shared hero. For each, give a concise title and a concise description (what to draw, the labels, which element takes the accent, the takeaway). Output as a table:

#	Section	Title	Description
0	Hero (1.91:1)	…	The whole thesis at a glance. Triples as Substack hero, Substack thumbnail, and the LinkedIn link-preview when cross-posted.
1	(first section)	…	…
2	…	…	…

• The hero is 1.91:1; inline section images are 16:9.
• At least one image should be a self-contained "mental model" diagram — the core idea drawn so it's screenshot-worthy on its own (that's how the piece gets pulled back onto LinkedIn/X by others). For a contrast, spec both states being compared, the labels, and the caption.
• Note placement: Substack images are uploaded manually at each section; Medium images carry over when the rendered HTML is copy-pasted (paths resolving next to the file).

Part B — LinkedIn: slide-by-slide PDF brief

LinkedIn document posts are PDFs, so spec the carousel slide by slide:

• Format — portrait 4:5 (1080×1350) or square 1:1; one accent reserved for the story's hero element; build slides share a motif; the turn/payoff slide breaks that rhythm; readable on a phone.
• Per slide — give the slide's role (hook / build / insight / turn / mental-model / where-it-goes / CTA), the exact text, and a one-line visual direction.
• Export — assemble the slides and export as a single PDF for the document post.

Always include a "what NOT to do" list

• Don't recycle the LinkedIn carousel slides into the Substack/Medium body — let each format stay native.
• Don't make the Substack post image-heavy (email rendering + deliverability).
• Don't use warning/error aesthetics for a positive-framed idea.
• Don't drift the running example — the same example shown in the post must appear in the visuals.