mk-youtube-get-caption

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

YouTube Caption Download

YouTube字幕下载

Download video subtitles and display content. Automatically detects video's original language and distinguishes between manual and auto-generated subtitles.

下载视频字幕并展示内容。自动检测视频的原始语言，区分人工上传字幕和自动生成字幕。

Quick Start

快速开始

/mk-youtube-get-caption <URL> [language|auto] [--force]

/mk-youtube-get-caption <URL> [language|auto] [--force]

Examples

示例

/mk-youtube-get-caption https://youtube.com/watch?v=xxx

- Auto-detect original language

/mk-youtube-get-caption https://youtube.com/watch?v=xxx auto

- Explicitly use original language

/mk-youtube-get-caption https://youtube.com/watch?v=xxx ja

- Download Japanese subtitles

/mk-youtube-get-caption https://youtube.com/watch?v=xxx "zh-TW,en"

- Language priority list

/mk-youtube-get-caption https://youtube.com/watch?v=xxx

- 自动检测原始语言

/mk-youtube-get-caption https://youtube.com/watch?v=xxx auto

- 明确使用原始语言

/mk-youtube-get-caption https://youtube.com/watch?v=xxx ja

- 下载日语字幕

/mk-youtube-get-caption https://youtube.com/watch?v=xxx \"zh-TW,en\"

- 语言优先级列表

Parameters

参数

Parameter	Required	Default	Description
URL	Yes	-	YouTube video URL
language	No	auto	Language code or priority list
--force	No	false	Force re-download even if cached file exists

参数	是否必填	默认值	说明
URL	是	-	YouTube视频URL
language	否	auto	语言代码或优先级列表
--force	否	false	即使存在缓存文件，强制重新下载

Language Options

语言选项

Value	Behavior
(empty)	Auto-detect video's original language
`auto`	Same as empty
`en` , `ja` , etc.	Specific language code
`"en,ja,zh-TW"`	Comma-separated priority list

值	行为
(空值)	自动检测视频的原始语言
`auto`	与空值行为相同
`en` , `ja` 等	指定语言代码
`\"en,ja,zh-TW\"`	逗号分隔的优先级列表

How it Works

工作原理

Execute:

{baseDir}/scripts/caption.sh "<URL>" "<language>"

If no language specified, detect video's original language
Try to download manual (author-uploaded) subtitles first
If unavailable, fall back to auto-generated subtitles
Parse JSON output to get file path and metadata
Use Read tool to get subtitle content if needed

执行：

{baseDir}/scripts/caption.sh \"<URL>\" \"<language>\"

如果未指定语言，检测视频的原始语言
优先尝试下载人工上传（作者上传）的字幕
如果人工字幕不可用，回退到自动生成的字幕
解析JSON输出以获取文件路径和元数据
如需获取字幕内容，使用Read工具读取

Output Format

输出格式

Success:

json

{
  "status": "success",
  "file_path": "{baseDir}/data/20091025__VIDEO_ID.en.srt",
  "text_file_path": "{baseDir}/data/20091025__VIDEO_ID.en.txt",
  "language": "en",
  "subtitle_type": "manual",
  "char_count": 30287,
  "line_count": 1555,
  "text_char_count": 25000,
  "text_line_count": 800,
  "cached": false,
  "video_id": "dQw4w9WgXcQ",
  "title": "Video Title",
  "channel": "Channel Name",
  "url": "https://www.youtube.com/watch?v=dQw4w9WgXcQ"
}

Cache hit (returns existing file):

json

{
  "status": "success",
  "file_path": "{baseDir}/data/20091025__VIDEO_ID.en.srt",
  "subtitle_type": "cached",
  "cached": true,
  ...
}

Error:

json

{
  "status": "error",
  "message": "No subtitles found (this video may not have subtitles)"
}

成功时：

json

{
  \"status\": \"success\",
  \"file_path\": \"{baseDir}/data/20091025__VIDEO_ID.en.srt\",
  \"text_file_path\": \"{baseDir}/data/20091025__VIDEO_ID.en.txt\",
  \"language\": \"en\",
  \"subtitle_type\": \"manual\",
  \"char_count\": 30287,
  \"line_count\": 1555,
  \"text_char_count\": 25000,
  \"text_line_count\": 800,
  \"cached\": false,
  \"video_id\": \"dQw4w9WgXcQ\",
  \"title\": \"Video Title\",
  \"channel\": \"Channel Name\",
  \"url\": \"https://www.youtube.com/watch?v=dQw4w9WgXcQ\"
}

命中缓存（返回已存在的文件）：

json

{
  \"status\": \"success\",
  \"file_path\": \"{baseDir}/data/20091025__VIDEO_ID.en.srt\",
  \"subtitle_type\": \"cached\",
  \"cached\": true,
  ...
}

错误时：

json

{
  \"status\": \"error\",
  \"message\": \"No subtitles found (this video may not have subtitles)\"
}

Output Fields

输出字段

Field	Description
`file_path`	Absolute path to the downloaded SRT file
`text_file_path`	Absolute path to plain text file (no timestamps)
`language`	Detected language code of the downloaded subtitle
`subtitle_type`	`manual` (author-uploaded) or `auto-generated` (YouTube AI)
`char_count`	Number of characters in the SRT file
`line_count`	Number of lines in the SRT file
`text_char_count`	Number of characters in the plain text file
`text_line_count`	Number of lines in the plain text file
`video_id`	YouTube video ID
`title`	Video title
`channel`	Channel name
`url`	Full video URL

字段	说明
`file_path`	下载的SRT文件的绝对路径
`text_file_path`	纯文本文件的绝对路径（无时间戳）
`language`	下载字幕的检测语言代码
`subtitle_type`	`manual` （作者上传）或 `auto-generated` （YouTube AI生成）
`char_count`	SRT文件中的字符数
`line_count`	SRT文件中的行数
`text_char_count`	纯文本文件中的字符数
`text_line_count`	纯文本文件中的行数
`video_id`	YouTube视频ID
`title`	视频标题
`channel`	频道名称
`url`	完整视频URL

Filename Format

文件名格式

Files use unified naming with date prefix:

{YYYYMMDD}__{video_id}.{lang}.{ext}

Example:

20091025__dQw4w9WgXcQ.en.srt

文件采用统一命名格式，带有日期前缀：

{YYYYMMDD}__{video_id}.{lang}.{ext}

示例：

20091025__dQw4w9WgXcQ.en.srt

Notes

注意事项

File caching: If caption file already exists for this video/language, it will be reused (returns
```
cached: true
```
)
Force refresh: Use
```
--force
```
flag to re-download even if cached file exists
Uses system yt-dlp/jq if available, otherwise auto-downloads on first run
Some videos may not have subtitles
Manual subtitles are prioritized over auto-generated ones
Auto-generated subtitles may be less accurate

文件缓存：如果该视频/语言的字幕文件已存在，将直接复用（返回
```
cached: true
```
）
强制刷新：使用
```
--force
```
参数可忽略缓存文件，强制重新下载
如果系统中存在yt-dlp/jq则直接使用，否则首次运行时会自动下载
部分视频可能没有字幕
优先使用人工上传的字幕，而非自动生成字幕
自动生成的字幕可能准确性较低

Next Step

下一步操作

After downloading the caption, invoke

/mk-youtube-transcript-summarize

with the

text_file_path

from the output to generate a structured summary:

/mk-youtube-transcript-summarize <text_file_path>

IMPORTANT: Always use the Skill tool to invoke

/mk-youtube-transcript-summarize

. Do NOT generate summaries directly without loading the skill — it contains critical rules for compression ratio, section structure, data preservation, and language handling.

下载字幕完成后，可调用

/mk-youtube-transcript-summarize

工具并传入输出中的

text_file_path

以生成结构化摘要：

/mk-youtube-transcript-summarize <text_file_path>

重要提示：请务必通过Skill工具调用

/mk-youtube-transcript-summarize

。请勿直接生成摘要而不加载该Skill——它包含了压缩比例、章节结构、数据保留和语言处理的关键规则。