Clean up Chinese AI-generated writing so it reads more like a human author wrote it.
Pull the full text out of a WeChat public account article without an API key.
Generate a motion-graphics video concept or a dark cinematic SaaS product video from a topic.
Check how closely a replica video matches an authorized reference video.
| pluviobyte/rnskill | robbyant/lingbot-vla-v2 | xiejunjie524/handdraw-story-video | |
|---|---|---|---|
| Stars | 613 | 613 | 616 |
| Language | Python | Python | Python |
| Setup difficulty | easy | hard | hard |
| Complexity | 2/5 | — | 3/5 |
| Audience | writer | researcher | designer |
Figures from each repo's GitHub metadata at analysis time.
Licensed CC BY-NC 4.0, so it cannot be used commercially without the author's permission.
rnskill is a collection of AI agent skills, maintained by a creator known as Xuetawuyun, built for Codex, Claude Code, and other AI agent tools that support the SKILL.md format. Each skill is a self-contained folder you add to a project so an agent can call on it for a specific task. The skills fall into four groups. One handles Chinese-language writing: an editor that removes patterns that read as AI-generated while keeping the original author's voice, facts, and judgment intact. Another extracts the full text of WeChat public account articles by spoofing the WeChat in-app browser, using only Python's standard library and no API key. Four skills are for video production: a motion-first video director that turns a topic into a video concept with visual metaphors and a beat-by-beat plan, a tool that builds an original, editable motion study based on an authorized reference video, complete with supporting analysis and a quality check on the final file, a dark, cinematic style generator for SaaS product videos with eight scene templates and three timing presets, and a black-and-white typed-text opener animation with synced typing sound effects, including a Python script that generates the timing plan. The last skill is a quality control checklist for judging how faithfully a replica video matches its reference, covering five fidelity levels plus checks on assets, runtime, and delivery. You can install the whole collection or a single skill through the Claude Code plugin marketplace, a universal command line installer that works for both Codex and Claude Code, or by manually copying a skill folder into your project's skills directory. Four of the video skills are actually maintained in a separate repository and mirrored here with a sync script. The project is released under the CC BY-NC 4.0 license, which does not permit commercial use.
A library of installable AI agent skills for Chinese writing cleanup, WeChat article extraction, and several styles of AI-directed video production, for use with Claude Code or Codex.
Mainly Python. The stack also includes Python, SKILL.md, Claude Code.
You may use and share this for non-commercial purposes only, and must credit the author, commercial use is not allowed.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly writer.
This repo across BitVibe Labs
Don't trust strangers blindly. Verify against the repo.