youtube-research

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

YouTube Research

YouTube研究

Follow shared release-shell rules in:

```
postplus-shared
```
release-shell rules

Use this skill when the request is about YouTube channels, audience proxies, or public videos.

This skill is for:

collecting channel metadata and subscriber counts
collecting audience comment samples from a specific video
collecting one or more public YouTube video URLs
discovering a small YouTube video set from a broad query
normalizing YouTube video metrics into local artifacts
writing local raw payloads, normalized datasets, and short markdown summaries

This skill is not for:

LinkedIn collection
Facebook collection
TikTok, Instagram, or X workflows
publishing, deleting, or scheduling posts
promising a subscriber list that YouTube does not expose publicly

遵循

postplus-shared

发布shell中的共享规则：

```
postplus-shared
```
release-shell规则

当请求涉及YouTube频道、观众代理或公开视频时使用此技能。

此技能适用于：

收集频道元数据和订阅者数量
收集特定视频的观众评论样本
收集一个或多个公开YouTube视频URL
通过宽泛查询发现少量YouTube视频集
将YouTube视频指标标准化为本地工件
编写本地原始有效载荷、标准化数据集和简短Markdown摘要

此技能不适用于：

LinkedIn数据收集
Facebook数据收集
TikTok、Instagram或X平台工作流
发布、删除或排期帖子
承诺获取YouTube未公开的订阅者列表

Hosted Capability Boundary

托管能力边界

This skill depends on host-managed collection capability for the corresponding collection keys.

In the product shell:

do not probe or print provider secrets
do not ask the user to export them inside chat
if a collection key returns a stable capability/network hard error, stop immediately instead of trying alternate shell commands

此技能依赖于主机管理的收集能力以及对应的收集密钥。

在产品shell中：

不要探查或打印提供商密钥
不要要求用户在聊天中导出密钥
如果收集密钥返回稳定的能力/网络硬错误，请立即停止，不要尝试替代shell命令

Default Collection Keys

默认收集密钥

Use these hosted collection keys by default:

channel metadata and subscriber counts

collection key:
```
youtube-channel-summary
```
use for channel metadata and subscriber counts

tested input shape:

```
channels = ["@Google"]
```
```
maxVideosPerChannel = 0
```
```
includeChannelInfo = true
```
```
includeVideos = false
```

current observed output fields include:

```
channelId
```
```
channelName
```
```
channelHandle
```
```
subscriberCount
```
```
totalVideos
```
```
avatarUrl
```
```
bannerUrl
```

audience comments collection
- collection key:
```
youtube-comments
```
- use for audience proxy research from comments on a specific video or Shorts URL
- tested input shape:
  - ```
  startUrls = ["https://www.youtube.com/watch?v=<video-id>"]
```
- ```
sort
```
  - ```
  maxItems
```
- current observed output fields include:
  - ```
  id
```
- ```
text
```
  - ```
  likeCount
```
- ```
replyCount
```
  - ```
  publishedTime
```
- ```
author
```
downloadable video records
- collection key:
```
youtube-video-download
```
- use when the workflow needs the rented downloader-backed record for explicit video URLs

Use comments as an audience proxy when subscriber identities are not public.

默认使用以下托管收集密钥：

频道元数据和订阅者数量

收集密钥：
```
youtube-channel-summary
```
用于频道元数据和订阅者数量

测试输入格式：

```
channels = ["@Google"]
```
```
maxVideosPerChannel = 0
```
```
includeChannelInfo = true
```
```
includeVideos = false
```

当前观测到的输出字段包括：

```
channelId
```
```
channelName
```
```
channelHandle
```
```
subscriberCount
```
```
totalVideos
```
```
avatarUrl
```
```
bannerUrl
```

观众评论收集
- 收集密钥：
```
youtube-comments
```
- 用于从特定视频或Shorts URL的评论中进行观众代理研究
- 测试输入格式：
  - ```
  startUrls = ["https://www.youtube.com/watch?v=<video-id>"]
```
- ```
sort
```
  - ```
  maxItems
```
- 当前观测到的输出字段包括：
  - ```
  id
```
- ```
text
```
  - ```
  likeCount
```
- ```
replyCount
```
  - ```
  publishedTime
```
- ```
author
```
可下载视频记录
- 收集密钥：
```
youtube-video-download
```
- 当工作流需要基于租用下载器的记录来处理明确的视频URL时使用

当订阅者身份不公开时，将评论用作观众代理。

Failure Posture

失败处理机制

fail if the request includes non-YouTube platforms
fail if no YouTube public URLs can be discovered
fail if hosted content collection returns malformed items without stable URL or id fields
fail if the channel collection cannot resolve the channel handle or URL
keep raw responses for debugging

如果请求包含非YouTube平台，直接失败
如果无法发现任何YouTube公开URL，直接失败
如果托管内容收集返回的项目格式错误，且缺少稳定的URL或id字段，直接失败
如果频道收集无法解析频道句柄或URL，直接失败
保留原始响应用于调试

Recommended Workflow

Release-Shell Execution Contract

发布Shell执行约定

keep collection briefs, raw datasets, normalized outputs, and summary caches under
```
<work-folder>/.postplus/youtube-research/
```
keep only final user-facing summaries or shortlist exports outside
```
.postplus/
```
compile a small channel, comment, or public-video brief before the expensive collection step
start with a bounded first pass:
- one channel batch
- one comments batch
- one public video plan

将收集简报、原始数据集、标准化输出和摘要缓存保存在
```
<work-folder>/.postplus/youtube-research/
```
目录下
仅将最终面向用户的摘要或候选列表导出到
```
.postplus/
```
目录外
在执行高成本的收集步骤之前，先编写一份简短的频道、评论或公开视频简报
从有限的首次执行开始：
- 一个频道批次
- 一个评论批次
- 一个公开视频计划

Public Video Implementation

公开视频实现

Read before implementation:

skills/20-research/youtube-research/references/normalized-schema.md

Use these entrypoints:

skills/20-research/youtube-research/scripts/run_youtube_video_collection.mjs

实现前请阅读：

skills/20-research/youtube-research/references/normalized-schema.md

使用以下入口点：

skills/20-research/youtube-research/scripts/run_youtube_video_collection.mjs

Hosted Collection Note

托管收集说明

Use the shared hosted collection runner for actor calls:

${CLAUDE_SKILL_DIR}/_postplus_shared/00-core/shared-collection/scripts/collection_actor_run.mjs

使用共享托管收集运行器进行actor调用：

${CLAUDE_SKILL_DIR}/_postplus_shared/00-core/shared-collection/scripts/collection_actor_run.mjs