官方说明开源项目5 天前
default-cybe/tech-explainer-video-skill
Make high-impact, premium-minimal animated tech-explainer videos locally (no paid APIs) — a Claude Code skill…
在 GitHub 查看 ↗直链 · 无中转页
官方说明Make high-impact, premium-minimal animated tech-explainer videos locally (no paid APIs) — a Claude Code skill + manim/Kokoro/faster-whisper engine.
- ★ 数取值
- 0 ★ · default-cybe/tech-explainer-video-skill · 取自 GitHub 搜索 API · （2026-10-16 20:00:20 核对）
- 来源明细
- 从该仓库的 GitHub 页面直接读取 · 条目类型：开源项目
- 收录与活跃
- 收录于 2026-10-12 · 最近见到 2026-10-16 · 共见到 7 次 · 被 1 个来源列出
- 评分构成
- 综合分 103 · 1 个来源 · repo
- 安装
pip install manim
常见问题（答案都来自我们的记录）
怎么安装？
该仓库给出的安装命令是
pip install manim。我们的记录里它是仓库自己提供的；不保证在你的客户端上直接可用。它适合谁 / 用来做什么？
我们把它归在这些标签下: 开源项目。标签来自条目自身，不是我们强加的分类。
谁维护它、还在更新吗？
仓库
default-cybe/tech-explainer-video-skill；我们收录于 2026-10-12；最近见到 2026-10-16。这里没有"官方认证"或"质量评级"——我们只记录来源与时间。第三方质量标记
tech-explainer-video-skill
以下摘录自该仓库的 README（片段） · 在 GitHub 读完整 README →
A Claude Code skill for making animated explainer videos about technical stuff: a bug, a system design
question, an algorithm. Everything runs on your own machine. No paid APIs, no API keys.
I built this because I wanted to actually understand my own security findings, and reading a writeup
three times wasn't doing it. A short animated walkthrough with narration did.
The visual style is minimal on purpose. Black background, white and grey text, one accent colour, and
one diagram that stays on screen and changes as the explanation moves along. If you've seen the system
design reels from Engineering Digest, it's that kind of look.
What's in here
- skills/impactful-explainer-video/ is the skill itself. Claude reads SKILL.md and follows it.
- engine/ has the Python pieces the skill uses:
- theme3.py for the visual style and reusable parts (cards, status pill, dotted cube, mode cells)
- tts.py for narration with Kokoro, a local text to speech model
- captions.py for word level captions using faster-whisper
- sfx.py for sound effects and a quiet music bed, generated in code
- example/scenes.py is a two scene demo you can render to check your setup.
Install the tools
You need Python 3.12 (3.14 is too new for some of the packages), ffmpeg, and the fonts JetBrains Mono
and Noto Sans.
Then download the Kokoro voice model once:
Add the skill to Claude Code
As a plugin:
On Claude Code 2.1.275 or newer you can do it in one line:
Or copy the folder in by hand:
After that, ask Claude something like "make an animated explainer video about how a token bucket rate
limiter works" and it will follow the skill.
How a video gets made
1. Write the script as a list of scenes, each with a short narration.
2. tts.py turns each narration into a wav file and records how long it is.
3. Each scene is a manim animation that runs as long as its narration.
4. captions.py gets word timings from the audio and writes karaoke style captions.
5. sfx.py places small sound effects where the animation logged them, plus a soft music bed.
6. ffmpeg puts it together: video, narration, sound, and burned in captions.
The exact ffmpeg commands are in skills/impactful-explainer-video/references/pipeline.md.
A few things I learned the hard way
- Saturated red or green text on pure black gets ugly speckles after H.264 compression. Use the
softer, brighter versions of those colours, render at 2x and scale down.
question, an algorithm. Everything runs on your own machine. No paid APIs, no API keys.
I built this because I wanted to actually understand my own security findings, and reading a writeup
three times wasn't doing it. A short animated walkthrough with narration did.
The visual style is minimal on purpose. Black background, white and grey text, one accent colour, and
one diagram that stays on screen and changes as the explanation moves along. If you've seen the system
design reels from Engineering Digest, it's that kind of look.
What's in here
- skills/impactful-explainer-video/ is the skill itself. Claude reads SKILL.md and follows it.
- engine/ has the Python pieces the skill uses:
- theme3.py for the visual style and reusable parts (cards, status pill, dotted cube, mode cells)
- tts.py for narration with Kokoro, a local text to speech model
- captions.py for word level captions using faster-whisper
- sfx.py for sound effects and a quiet music bed, generated in code
- example/scenes.py is a two scene demo you can render to check your setup.
Install the tools
You need Python 3.12 (3.14 is too new for some of the packages), ffmpeg, and the fonts JetBrains Mono
and Noto Sans.
Then download the Kokoro voice model once:
Add the skill to Claude Code
As a plugin:
On Claude Code 2.1.275 or newer you can do it in one line:
Or copy the folder in by hand:
After that, ask Claude something like "make an animated explainer video about how a token bucket rate
limiter works" and it will follow the skill.
How a video gets made
1. Write the script as a list of scenes, each with a short narration.
2. tts.py turns each narration into a wav file and records how long it is.
3. Each scene is a manim animation that runs as long as its narration.
4. captions.py gets word timings from the audio and writes karaoke style captions.
5. sfx.py places small sound effects where the animation logged them, plus a soft music bed.
6. ffmpeg puts it together: video, narration, sound, and burned in captions.
The exact ffmpeg commands are in skills/impactful-explainer-video/references/pipeline.md.
A few things I learned the hard way
- Saturated red or green text on pure black gets ugly speckles after H.264 compression. Use the
softer, brighter versions of those colours, render at 2x and scale down.
- 来源
- github · https://github.com/default-cybe/tech-explainer-video-skill
- 按领域浏览
- 开源项目
- 收录于
- 昨天
更多开源项目
所属领域与标签
每个标签都能点进去看同类的全部条目。数字是当前库里的真实条数。