knowledge-video skill
A skill for knowledge explainers: the agent researches a topic, writes screen copy and builds a HyperFrames project for MP4
- 61 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The skill searches the web, downloads images and writes project files with a render.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/knowledge-videoDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The skill searches the web, downloads images and writes project files with a render.
This is third-party code. Review the repository files before installing.
What it does
`knowledge-video` takes an agent from a topic to a two or three minute knowledge explainer in the spirit of the Douyin Vibe Knowledge tag: animation drawn in code with short on-screen lines over music. The agent does not generate video; it researches the topic, writes copy and builds a HyperFrames project rendered to MP4. The path has six steps: sourced research (a fact without a reliable source is cut), screen copy up to 20 characters, image picks with licenses, look and storyboard, build and render. After each step the agent shows the result so a wrong direction is caught before a long render. Four hard rules apply: every fact has a source, every asset a license, no AI-drawn faces of real people, and frames render the same every time.
Who it is for. Creators of educational clips and teachers who need a verifiable treatment of a topic with sources and image licenses.
Good fit when
- A topic such as why the sky is blue should become a short animated explainer
- The clip needs a sourced article and image credits
- A historical overview is built as a timeline with real photos and documents
- Music is fitted to the beat of screen changes and sound effects are added per frame
Not a fit when
- You need video with real people or footage
- The machine has no Python with `numpy` and `FFmpeg`
- The clip would be published without sources and AI labels
Example request
Make a two-minute explainer on why the sky is blue: sourced research first, then copy by screen.Limitations
It needs Python 3 with `numpy`, `FFmpeg` and a HyperFrames build. The skill gives a path and small tools while leaving copy, look and motion to the agent, so results depend on the model. Images are picked with licenses (CC0 and public domain first), but rights and attribution remain your responsibility. Before publishing the README advises adding sources to the description and turning on the AI-content label. The main examples are Chinese, with no Russian ones.
How to disable. Delete the `knowledge-video` folder from `~/.claude/skills` or run `/plugin uninstall knowledge-video`. Finished projects and clips live in your directory and are removed separately.
FAQ
How do I install it?
Tell the agent to install the skill from `https://github.com/Win-Hao/knowledge-video`, run `npx skills add Win-Hao/knowledge-video -g`, or add the plugin with `/plugin marketplace add Win-Hao/knowledge-video`.
What does the agent do first?
It picks the shape of the topic (timeline, chain of causes, shallow to deep), searches the web and writes a sourced article, dropping facts with no reliable support. You choose the number of points.
Does it generate video with a model?
No, frames are drawn in code in a HyperFrames project and rendered to MP4. The model writes copy and code, not images.
What should I do before publishing?
Put sources and credits in the description (the skill writes `credits.md`), mark modified images and turn on the AI-content label if the platform has one.
Does it work in Cursor and Codex?
The README says any `SKILL.md` agent works, and the `npx skills` install or manual copy from `skills/` applies there too.
Discussion
No questions or reviews yet.
Related
SkillsA skill that turns an idea, a link or a screen recording into finished video: HeyGen avatar, b-roll, captions and quality checks
Skills for faceless video channels: one topic becomes a video with voiceover, subtitles, covers and per-platform copy.
A skill turns a topic or script into a drawn explainer through a brief, storyboard, sample segment and local rendering.
A Claude Code and Codex skill that turns a topic into a narrated motion-graphics explainer video on Remotion with subtitles and chapters
The Video by Reference skill learns style from a sample clip and builds an animated short from code with an original soundtrack
A skill that turns a script into a finished video with voice-over, subtitles, a cover and a publish text, frames drawn by code.
README in short
The documentation explains the idea, six steps with what the human checks, requirements, install routes, pre-publishing rules and repository layout. A Chinese README is included.