liucongg/liucong-skills — explained in plain English
Analysis updated 2026-05-18
Generate a full storyboard and video prompt package from one sentence for a short 3D-to-2D style action film.
Turn a WeChat article draft into several ranked title options across different styles with risk notes.
Review and critique an existing set of article titles using the title-strategist skill's diagnostic process.
| liucongg/liucong-skills | binghe1980/ai-canvas | duolahypercho/codex-router | |
|---|---|---|---|
| Stars | 182 | 182 | 181 |
| Language | — | TypeScript | JavaScript |
| Setup difficulty | easy | moderate | moderate |
| Complexity | 1/5 | 2/5 | 4/5 |
| Audience | writer | designer | developer |
Figures from each repo's GitHub metadata at analysis time.
Install a single skill folder via symlink or copy into your agent platform's skills directory.
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.
liucong-skills is a small library of two AI agent skills: one turns a sentence into a full video-prompt package for 3D-to-2D short films, the other rewrites and ranks WeChat article titles.
Apache License 2.0: use, modify, and redistribute freely, including commercially, as long as you keep the license and any notice files.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly writer.
This repo across BitVibe Labs
Verify against the repo before relying on details.