detail-flow

Plan, generate, audit, and deliver multi-screen ecommerce detail pages from product images.

190|28|Updated Jun 11, 2026
One-click install
npx skills add https://github.com/AJbeckliy/detail-flow --skill detail-flow-ajbeckliy
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: detail-flow
Source: https://github.com/AJbeckliy/detail-flow
Command: npx skills add https://github.com/AJbeckliy/detail-flow --skill detail-flow-ajbeckliy

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Creating a coherent multi-screen ecommerce detail page from product photos is error-prone: screens drift in style, copy repeats the same headline structure, and AI image generation invents unsupported claims. This Skill enforces a structured blueprint-first workflow with approval gates, visual masters, staged previews, and targeted revision routing. ## Core Features & Use Cases - Blueprint-first planning: Outputs an 8-screen page blueprint with buyer questions, claim seeds, copy structure patterns, and edge anchors before any image generation, requiring user approval at two defined gates. - Visual continuity system: Builds a text visual master and an optional 1:3 image master to lock palette, lighting, typography, and product identity across all 9:21 slices, with early 2-slice concat previews to catch failures. - Audit and minimal revision routing: Checks for product drift, garbled text, unsupported claims, and standalone-poster repetition, then regenerates only the smallest affected layer instead of the full set. - Use Case: Provide a headphone product photo and a style reference image, confirm the generated 8-screen plan, and receive a delivery folder containing the master, individual slices, previews, and a full long-image concat. ## Quick Start Provide a product image and a style reference image, then ask the assistant to generate an 8-screen ecommerce detail page for the product.

Frequently Asked Questions about detail-flow

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate an 8-screen ecommerce detail page from a product image?

Provide a product image and optionally a style reference image, then request an 8-screen detail page. The workflow first outputs a structured blueprint for approval, then builds visual masters, generates 9:21 slices in stages, and delivers a folder with slices, previews, and a full long image.

How do I keep AI-generated product page screens visually consistent?

Use a text visual master plus an optional 1:3 image master to lock palette, lighting, typography, and product identity before generating slices. Staged concat previews after the first two slices catch continuity failures early, and edge anchors define what carries between adjacent screens.

Can I generate a detail page without providing product selling points?

Yes, the workflow infers selling points from the product image and category when none are supplied. However, AI-inferred claims are flagged for review before commercial use, and exact parameters, certifications, or awards are never invented.

What happens when a generated screen has product drift or garbled text?

The audit step routes the fix to the smallest responsible layer instead of regenerating everything. Copy problems revise only that slice's text, product drift strengthens reference constraints for the affected slice, and transition issues regenerate only the neighboring slice range.

What are the limitations of AI-generated ecommerce detail pages?

Image models can drift on product structure and text details, so human review remains part of the process. Independently generated slices cannot guarantee seamless joins like a single photo, and medical, certification, or precise parameter claims must come from the user.