wtf is liucong-skills?

liucong-skills is a small collection of agent skills, each living in its own folder under a skills directory and installable one at a time into whatever AI agent or platform you use. Right now it holds two skills. The first, painterly-3d2-cinema, turns a single sentence description into a full package of prompts for a short 3D-to-2D style action video. It walks through offering three story direction options, locking in a plot broken into 15 second segments, building character and scene prompts, producing a 15 panel storyboard prompt for Midjourney V8.1 for each segment, and finally generating matching video prompts for a tool called Seedance. It is worth noting clearly that this skill only produces text prompts. It does not generate any images or video itself, those still need to be created using Midjourney and Seedance separately and then fed back to the agent for review. The second skill, gzh-title-strategist, is built for people writing WeChat official account articles. Given an article, a draft, or an existing title, it first works out the article's type, target reader, core conclusion, and supporting evidence, then generates and rewrites titles across several styles such as steady and accurate, internet-catchy, authoritative, and data or keyword focused. It scores the top candidates and returns one main recommendation, two backup titles, and a note on any risks, including checks for clickbait phrasing and keyword stuffing. The README specifies that it will only use historical data actually provided or built into the skill, and will not invent numbers, companies, or claims that are not backed by the article's own text. Both skills install the same way, either as a symlink so repository updates apply immediately, or as a plain copy that needs to be redone manually after each update. The project is released under the Apache License 2.0.