Youtube Transcriber Skill

自动转录 YouTube 视频，生成带时间戳的文字稿

Security Scan

The skill appears to do what it claims (download a YouTube video's audio and transcribe it locally with faster-whisper) and its files/requirements are proportionate to that purpose.

Purpose & Capability

Name/description match the code and docs: the package uses yt-dlp to download audio and faster-whisper to transcribe. Required packages and entrypoints (transcribe.py, bin/transcribe.js) align with the stated purpose; no unrelated credentials or binaries are requested.

Instruction Scope

SKILL.md and the included scripts instruct the agent/user to download audio via yt-dlp and run local transcription. The runtime instructions reference only the YouTube URL, language option, and local commands; they do not read unrelated system files or request secrets.

Install Mechanism

There is no high-risk remote installer; installation is via pip install -r requirements.txt (install.sh). This is expected for a Python tool but will install packages system-wide unless run in a virtual environment. The transcription model (faster-whisper/WhisperModel) will likely download model weights from external model hosts at runtime, which implies large network and disk usage — expected but worth noting.

Credentials

The skill declares no required environment variables or credentials and the code does not access secrets. README mentions optional proxy env vars for network connectivity, which is reasonable and optional.

Persistence & Privilege

The skill is user-invocable and not forced always-on. The included openclaw/install metadata indicates an auto-install script (install.sh) will run when installing, which is normal for installing dependencies but does not modify other skills or system configurations beyond pip installs.

Assessment

This package appears coherent and implements local transcription: it downloads audio with yt-dlp and runs faster-whisper locally. Before installing, consider: (1) run install.sh or pip inside a virtualenv to avoid system-wide installs; (2) model weights will likely be downloaded at first run (large bandwidth and disk usage); (3) yt-dlp downloads YouTube content — ensure you have the right to download the videos you process; (4) faster-whisper may require sufficient RAM/CPU or GPU and ffmpeg might be needed for some formats. If you want extra safety, inspect/run the code in an isolated environment (container or VM) first.

自动转录 YouTube 视频，生成带时间戳的文字稿和结构化总结。

使用方法

# OpenClaw 中使用
/youtube-transcribe <URL> [language]

# 例子
/youtube-transcribe "https://youtube.com/watch?v=VIDEO_ID"
/youtube-transcribe "https://youtube.com/watch?v=VIDEO_ID" zh

参数

参数	说明	默认值
URL	YouTube 视频链接	必填
language	语言代码 (zh/en/ja/ko/auto)	auto

依赖

• Python >= 3.8.0
• yt-dlp
• faster-whisper

例子

# 自动检测语言
/youtube-transcribe "https://youtube.com/watch?v=jNQXAC9IVRw"

# 指定中文
/youtube-transcribe "https://youtube.com/watch?v=9uDH8z-HZKs" zh

# 指定英文
/youtube-transcribe "https://youtube.com/watch?v=VIDEO_ID" en

输出示例

📺 URL: https://youtube.com/watch?v=VIDEO_ID
🌐 Language: zh

[1/3] 下载音频...
✅ 下载成功

[2/3] 转录中...
✅ 转录完成！

[3/3] 生成总结...
============================================================
# 视频转录总结

**URL:** https://youtube.com/watch?v=VIDEO_ID
**语言:** zh

## 转录内容
[0.00s -> 3.00s] 大家好...
[3.00s -> 5.00s] 今天我们来...
============================================================