what is atutun-xhs-cover-v2 fr?

This repository is a skill definition meant to be plugged into an AI coding agent, most reliably Codex according to the author, though the author notes other agents such as WorkBuddy, Trae, and Kimi Work have not yet worked successfully with it. Its purpose is to turn written content such as articles, scripts, topic ideas, product reviews, AI tool tutorials, productivity tips, podcast recommendations, or personal experience posts into a detailed text prompt that can be fed directly into an image generation model to produce a cover image for Xiaohongshu, a popular Chinese social media platform. The project is on its second version, and the main improvement over the first version is a much shorter question flow. Instead of asking the user to manually choose every visual detail such as pose, props, background, font, and title color one by one, version two narrows this down to five essential questions: who the person in the cover is, what supporting materials to include, the title text, the overall cover style, and any preferred title color. Everything else, including pose, background, font, color palette, and layout elements, is filled in automatically by the agent based on the content and chosen style. The repository contains a main instruction file describing the five question flow, a review checklist run before the final prompt is produced, and eight named cover style templates, plus a small configuration file used specifically for the OpenAI or Codex side that sets a display name and description. To use it with Codex, you copy the skill folder into the local Codex skills directory and invoke it by name or by describing what kind of cover you need. Using it with other AI agents is described as only theoretically possible, by importing the instruction file as a system prompt or knowledge base entry, with no guarantee it behaves the same way. The repository is public and intended purely for sharing and reusing this workflow, and it does not include any personal files or private example images.