aplus-premium-build

name	aplus-premium-build
description	Full A+ Premium build pipeline — research brief → 7 module designs (Higgsfield) → 1 hero video (Higgsfield Veo/Sora) → comprehension test (ProductPinion video) → designer-ready output. Use when running `aplus-premium-build {ASIN}` for brands eligible for A+ Premium and willing to invest in cinema-quality assets.

aplus-premium-build — Full A+ Premium Pipeline

A+ Premium gives brands the highest-impact PDP real estate on Amazon. This is the end-to-end build.

Methodology — read before rendering

Read reference/02-visual-content/aplus-creative-director.md in full before generating any A+ module. (Updated in v0.1.7 — A+ has its own dedicated creative-director ref now, separate from listing-image.) That file is the source of truth for: the 13-step A+ flow, the 4 A+ Critical Rules, the 5 Pre-Generation Checklist questions, editorial design philosophy, anti-clipart rules, 6-section prompt structure (A+ variant), 4-axis scoring rubric, comparison-module exception, and A+ module taxonomy.

Hard rules:

• Aspect ratio: 16:9 widescreen on every module — locked at start AND end of every prompt
• Reference image always: live product photo passed to Higgsfield as image-to-image reference with every call
• Brand guidelines respected: from brain.business.brand_guidelines
• Stack consistency: score the SET (5-7 modules + hero video) as well as each module — visual cohesion is non-negotiable. Use Soul Character training first when available.
• Editorial photography only — no clipart. Comparison-module exception applies for genuine spec/size tables.

Prerequisites

• All 3 MCPs
• Brand qualifies for A+ Premium (Vine + ≥5 published A+ pages typically)

Invocation

aplus-premium-build {ASIN}
aplus-premium-build {ASIN} --modules=7 --include-video

Phases

Phase	Skill	Output
1	asin-deep-research (or pull existing)	Research brief
2	rufus-gap-analysis	FAQ source for bottom half
3	aplus-module-generator --premium --modules=7	7 modules, two-half architecture
4	hero-video-builder (optional but recommended)	Cinematic hero video
5	aplus-comprehension (mockup test)	Validate scannability
6	Designer brief + Brand Registry deploy spec

Output

/tmp/cro-content/{ASIN}-aplus-premium-{date}.md — full deck + assets directory.

Output Format

# A+ Premium — {ASIN}

**Date:** {date} | **Modules:** {N} | **Hero video:** {included / skipped}

## Module Set

[Two-half architecture per `02-visual-content/a-plus-content.md`]

### Top Half — Capture & Convert
1. Hero (with optional video) ![](module-1.png)
2. Benefit deep-dive ![](module-2.png)
3. Differentiation ![](module-3.png)

### Bottom Half — Inform & Reassure
4. FAQ #1 (Rufus-sourced) ![](module-4.png)
5. FAQ #2 ![](module-5.png)
6. Comparison ![](module-6.png)
7. Final reassurance ![](module-7.png)

## Hero Video

![Video preview](hero.gif) — {duration, model used}

## Comprehension Score

{score}/100 — {pass/fail}

## Designer Brief

- Specs per module (970×600 full-width / 487×487 two-up)
- Hero video specs (16:9, 1080p)
- Alt-text packed per module

## Brand Registry Deploy Spec

[Step-by-step spec to upload via Brand Registry A+ editor]

Reference Files

• Anthropic skill: amazon-aplus-brief (companion for client-ready Word doc)
• Vault: CRO-Knowledge-Base/02-visual-content/a-plus-content.md

Quality Bar

• All 7 modules built + comprehension-tested
• Two-half architecture preserved
• Hero video if --include-video
• Designer brief detailed enough to ship without further questions

v0.1.12 — Read brain.image_strategy before generating

Before drafting any Higgsfield prompt, read brain.image_strategy from working memory:

• If null → pause and offer image-strategy first (~10 min, makes every future render bespoke). If user proceeds without, flag concept cards with "No category research used — generic aesthetic."
• If set + fresh (<90d) → inject prompt_adjustments.scene_keywords into prompt section 1, palette_keywords + mood_keywords into section 4, and do NOT include {anti_patterns_csv} near the end. Cite the strategy on each concept card.
• If stale (>90d) → flag and recommend image-strategy --refresh.

This brand-level strategy comes from skills/image-strategy.md's top-15-bestsellers analysis. Same adjustments apply across every render for this brand.

v0.1.13 — Visual verification gate + base64 embedding (NON-NEGOTIABLE)

Florence does NOT present an image she hasn't actually looked at, AND she does NOT use raw Higgsfield URLs in artifact HTML. Two blocking rules added in v0.1.13:

Verification gate

After Higgsfield returns each generated image, BEFORE adding to any artifact OR sending to Pinion:

1. Load the image into Florence's multimodal context — paste the URL in chat so Cowork's multimodal Claude loads it natively. Narrate: "Looking at #{N} before I include it."
2. Visually inspect against: aspect ratio (1:1 main+listing / 16:9 A+), product fidelity vs reference, technique landed (visibly), background appropriate, no clipart leak (anti-clipart rules), no text-on-main-image (TOS), no model faces.
3. If anything fails → re-prompt with specific fix + regenerate. Cap at 3 attempts per concept; surface honestly on attempt 4.
4. Only after ALL concepts pass → proceed to artifact emission.

The eye trumps the score. Even if the rubric said 100/100, if visual inspection finds a wrong product or off-aspect output, it fails the gate.

Base64 embedding

Higgsfield URLs are temporary AND Cowork's artifact iframe sandbox blocks external image loads in many builds. For every verified concept:

1. HTTP GET the Higgsfield URL → fetch image bytes
2. Detect MIME type from response headers (typically image/png)
3. Base64-encode the bytes
4. Substitute {{image-src}} (or {{winner-image-src}} for tests) with data:image/png;base64,<encoded> — NOT the raw Higgsfield URL

This makes the artifact self-contained — survives sandbox + URL expiry. ~1-3 MB per image is fine for Cowork.

The live product image ({{product-image-url}} in concepts.html hero strip / dossier head / cockpit product cards) stays as the live Amazon CDN URL — that's permanent and not affected.