YouTube Clipper Skill
A Claude Code skill that downloads a YouTube video, splits it into semantic chapters, cuts chosen clips, translates subtitles and burns bilingual ones in
- 2,254 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The scripts download videos from YouTube through `yt-dlp` and write files to the working folder
- The install script installs Python packages and creates an `.env` file in the skill folder
- Subtitle text is passed to the agent's model for analysis and translation
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/youtube-clipper-skillDetects 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 scripts download videos from YouTube through `yt-dlp` and write files to the working folder
- The install script installs Python packages and creates an `.env` file in the skill folder
- Subtitle text is passed to the agent's model for analysis and translation
This is third-party code. Review the repository files before installing.
What it does
YouTube Clipper Skill cuts a YouTube video into clips by semantic chapters and works in six stages. First the skill checks the environment: `yt-dlp`, FFmpeg with libass support and the Python packages `pysrt` and `python-dotenv`. Then, given a link, `download_video.py` uses `yt-dlp` to fetch the video at up to 1080p as mp4 and English VTT subtitles, falling back to automatic ones when none exist. The key step is done by the agent itself: it reads the whole subtitle track, finds topic changes and splits the recording into chapters of 2 to 5 minutes instead of cutting mechanically by time. For each chapter it writes a title, a time range, a short summary and 3 to 5 keywords, then asks you to pick chapters and options: bilingual subtitles, burning into the video, a summary text for social networks. For every chosen chapter `clip_video.py` cuts the clip through FFmpeg, the subtitle lines for that span are taken from the full track with shifted timestamps, and the agent translates them in batches of 20 lines into the target language (Chinese by default). `merge_bilingual_subtitles.py` builds a bilingual SRT with English on top, `burn_subtitles.py` burns it into the clip in white text with a black outline, and `generate_summary.py` prepares a short text for Xiaohongshu, Douyin and WeChat. The result goes to `./youtube-clips/<date and time>/<chapter>/`: the plain clip, the version with subtitles, the bilingual SRT and the summary file.
Who it is for. For short-video creators and translators who cut long YouTube interviews and lectures into themed clips with bilingual subtitles and work in Claude Code.
Good fit when
- A one-hour interview needs to become short themed clips with boundaries set by meaning, not by a timer
- An English video needs bilingual subtitles burned into the picture and a separate SRT file
- Each clip needs a ready text for publishing on social networks
- A lecture or course has to be split into chapters with titles, summaries and keywords
- Results should sit in a clear folder and date structure, with processing done in a batch over the chosen chapters
Not a fit when
- The video is not from YouTube: the download script is built on `yt-dlp` and YouTube English subtitles
- The video has no English subtitles, not even automatic ones: chapters are built only from the subtitle text
- There is no FFmpeg with libass on the system: on macOS the standard Homebrew build cannot burn subtitles and `ffmpeg-full` is needed
- You do not work in Claude Code: the skill is built for it, and translation is done by the agent itself during the session
Example request
Split https://youtube.com/watch?v=VIDEO_ID into semantic chapters and, for the ones I choose, make bilingual subtitles and burn them into the clipLimitations
The skill is oriented to the English and Chinese pair: the `.env` settings file holds `TARGET_LANGUAGE=中文`, and the summary texts are meant for Chinese platforms. Another target language is a settings change, but translation quality depends on the model. Video is downloaded at up to 1080p by default (`MAX_VIDEO_HEIGHT`), and chapter length is set by `TARGET_CHAPTER_DURATION`.
Burning subtitles needs FFmpeg with libass; on macOS it has to be replaced with `ffmpeg-full`. Downloading and re-encoding large videos takes time and disk space. Downloading YouTube recordings is limited by the platform's rules and the authors' rights: use the skill for material you have the right to, and do not publish other people's recordings without permission. The latest project update came in January 2026. MIT license.
How to disable. Delete the `~/.claude/skills/youtube-clipper` folder. Finished clips live in `youtube-clips` in the folder you ran the skill from and are deleted separately. The Python packages installed by the install script can be removed with `pip uninstall yt-dlp pysrt python-dotenv`.
FAQ
How does the skill split the video into chapters?
The agent reads the whole subtitle track with timestamps, finds topic changes and builds chapters of 2 to 5 minutes, each with a title, time range, summary and keywords. `analyze_subtitles.py` only parses VTT and hands over the text, and the model decides the boundaries.
Who translates the subtitles?
The agent itself during the session: `translate_subtitles.py` prepares batches of 20 lines and prints the translation requirements (term accuracy, spoken style, brevity), and the model returns the translation. Batch size is set by `TRANSLATION_BATCH_SIZE`, and on failure the translation is retried up to three times.
Why can't I burn subtitles on macOS?
The standard Homebrew `ffmpeg` is built without the libass library, so it has no `subtitles` filter. Replace it with `brew install ffmpeg-full` after removing the regular one. To check, `ffmpeg -filters 2>&1 | grep subtitles` should print the filter line.
What is in the results folder?
`./youtube-clips/<date and time>/<chapter>/` holds `_clip.mp4` without subtitles, `_with_subtitles.mp4` with burned-in subtitles, `_bilingual.srt` and `_summary.md` with a text for publishing. File names are cleaned of the characters `/ \ : * ? " < > |`, spaces become underscores, and length is capped at 100 characters.
Which settings can I change?
In `~/.claude/skills/youtube-clipper/.env`: `FFMPEG_PATH`, `OUTPUT_DIR`, `MAX_VIDEO_HEIGHT` (720, 1080, 1440 or 2160), `TRANSLATION_BATCH_SIZE` (20 to 25 recommended), `TARGET_LANGUAGE`, `TARGET_CHAPTER_DURATION` (180 to 300 seconds), plus subtitle font size and margin.
Discussion
No questions or reviews yet.
Related
SkillsA skill and MCP server for cutting long videos: the agent takes a link or file and builds Shorts, Reels and TikTok clips with a cover, post text and ZIP
YouTube Skills
YouTube Skills for AI Agents
A set of skills that gives an agent YouTube transcripts, video and channel search, and playlist extraction over one API
The /watch skill turns a link or file into frames and captions, or with a Gemini key hands the whole video to Google for the agent to answer from
ChatGPT short-video editing skills
ChatGPT 剪短影音 Skills
Two agent skills: set up the editing environment and build a vertical short video in eight verified steps
ffmpeg-skill — a skill and MCP server: an agent cuts and joins video, fixes audio, adds captions and exports for TikTok and Reels with local FFmpeg
A skill that pulls highlights from a long video: download, transcript, segmentation, analysis and clips with titles
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| YouTube Clipper Skill (this entry) | no | 2,254 | ||
| video-use | no | 29,550 | edits a video from raw takes through conversation with the agent: cuts pauses, color-corrects and adds subtitles | |
| Violin | no | 1,083 | translates and dubs video: speech transcription, translation, voice synthesis and subtitles | |
| YouTube Transcript MCP Server | no | not required | 492 | only hands the agent the transcript of a YouTube video with timestamps and metadata, with no clipping |
README in short
The description in English and Chinese lists features, two ways to install, system requirements with a libass warning, the six-stage workflow and the output files. It continues with settings, two examples, a comparison of semantic against mechanical splitting and of batch translation, a troubleshooting section and links to documentation. The repository holds technical notes on FFmpeg and yt-dlp, a fixes changelog, subtitle references and a summary text template.
README badge
Original SKILL.md
---
name: youtube-clipper
description: >
YouTube 视频智能剪辑工具。下载视频和字幕，AI 分析生成精细章节（几分钟级别），
用户选择片段后自动剪辑、翻译字幕为中英双语、烧录字幕到视频，并生成总结文案。
使用场景：当用户需要剪辑 YouTube 视频、生成短视频片段、制作双语字幕版本时。
关键词：视频剪辑、YouTube、字幕翻译、双语字幕、视频下载、clip video
allowed-tools:
- Read
- Write
- Bash
- Glob
- AskUserQuestion
model: claude-sonnet-4-5-20250514
---
# YouTube 视频智能剪辑工具
> **Installation**: If you're installing this skill from GitHub, please refer to [README.md](README.md#installation) for installation instructions. The recommended method is `npx skills add https://github.com/op7418/Youtube-clipper-skill`.
## 工作流程
你将按照以下 6 个阶段执行 YouTube 视频剪辑任务：
### 阶段 1: 环境检测
**目标**: 确保所有必需工具和依赖都已安装
1. 检测 yt-dlp 是否可用
```bash
yt-dlp --version
```
2. 检测 FFmpeg 版本和 libass 支持
```bash
# 优先检查 ffmpeg-full（macOS）
/opt/homebrew/opt/ffmpeg-full/bin/ffmpeg -version
# 检查标准 FFmpeg
ffmpeg -version
# 验证 libass 支持（字幕烧录必需）
ffmpeg -filters 2>&1 | grep subtitles
```
3. 检测 Python 依赖
```bash
python3 -c "import yt_dlp; print('✅ yt-dlp available')"
python3 -c "import pysrt; print('✅ pysrt available')"
```