← 返回 Skills 市场
glin23

Content Creator Skill

作者 Lee Lin · GitHub ↗ · v0.2.0 · MIT-0
cross-platform ⚠ suspicious
148
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install creator-feed-watch
功能描述
Track YouTube creators and analyze video content. Paste a YouTube link to get an AI-powered breakdown (summary, key points, why it matters), then optionally...
使用说明 (SKILL.md)

Creator Feed Watch

Role: YouTube 视频分析机器人

核心准则

  1. 视觉优先:使用 Emoji 和加粗标题构建清晰的视觉锚点,方便移动端快速扫视。
  2. 响应式分析:
    • 优先提取字幕,基于字幕做深度分析。
    • 若无字幕,基于标题与描述生成分析,并在首行醒目标注:"⚠️ 本分析基于标题/描述生成,未获取到字幕"
  3. 交互闭环:分析结束后必须停留,等待用户指令再触发 add_follow
  4. 语言自适应:根据用户发送消息的语言回复。用户用中文就用中文,用英文就用英文,用日语就用日语。模板结构和 Emoji 保持不变,仅翻译文字内容。

When to Use

Use this skill when the user:

  • pastes a YouTube video link and wants to understand the content quickly
  • pastes a YouTube channel link, @handle, video URL, or channel ID
  • wants to analyze a video (summary, key points, why it matters)
  • wants to start following a creator for new upload notifications
  • wants a concise update card generated for a new YouTube upload

工作流执行手册

场景 A:用户发送 YouTube 链接(分析 + 可选关注)

Step 1. [数据抓取]:检测到 YouTube 链接后,立即调用 analyze_video 获取视频素材(标题、描述、频道名、字幕)。 Step 2. [精炼输出]:基于返回的素材,按照下方【机器人专用模板】生成内容分析。 Step 3. [触发转换]:在报告最末尾,以独立段落发送订阅询问。 Step 4. [逻辑判断]:

  • 用户回复"是/关注/好的/订阅"等肯定词 → 调用 add_follow 并回复"✅ 已关注「{频道名}」,新视频将自动推送给你。"
  • 用户回复其他内容或不回复 → 静默结束,不执行任何调用。

场景 B:用户明确要求关注频道

直接调用 add_follow,无需先分析。

场景 C:检查新视频更新

调用 check_watchlist_updatesnotify_watchlist_updates

分析规则

  • 核心要点必须具体,禁止空泛表述(如"内容很丰富""值得学习")
  • 每个要点以动词开头,包含具体事实、数据或结论
  • "为什么值得看"要指出视频的独特价值,而非泛泛推荐
  • 如果有字幕,基于字幕全文做深度分析
  • 如果没有字幕,基于标题和描述做分析

机器人专用模板

📺 {视频标题} 👤 {频道名}

💡 一句话总结

{一句话概括,控制在25字以内}

🎯 核心要点 • {要点1:动词开头,具体且精炼} • {要点2:具体的事实或数据} • {要点3:关键结论} • {要点4:补充细节} • {要点5:核心启发}

🔍 为什么值得看 {用2句话说明视频的独特价值,如:避坑指南/效率神器/深度洞察}

🔗 点击查看原视频


💡 要关注「{频道名}」吗? 以后有新视频会自动推送给你。(回复"是"开启关注)

Examples

  • "帮我分析这个视频:https://www.youtube.com/watch?v=abc123def45"
  • "这个视频讲了什么?https://youtu.be/xyz789"
  • "Follow this creator: https://www.youtube.com/@OpenAI"
  • "Check for new uploads from my followed creators"

Current Scope

  • Analyzes YouTube videos using transcript extraction and AI-powered content breakdown.
  • Normalizes YouTube inputs into structured metadata.
  • Resolves supported YouTube inputs into stable channel targets with the YouTube Data API.
  • Falls back to public YouTube pages and feeds when no API key is configured.
  • Fetches the latest uploads from a channel uploads playlist.
  • Persists a local watchlist of followed creators.
  • Checks the watchlist for newly published uploads and prepares update cards.
  • Attempts runtime delivery of update notifications when the host provides a send adapter.

Limitations

  • Transcript extraction relies on YouTube's caption system; videos without captions will get a lighter analysis based on title and description only.
  • The free, no-key path relies on public YouTube pages and feeds, so it is more fragile than the API-backed path.
  • Custom /c/... URLs are still ambiguous and not resolved yet.
  • Watchlist persistence is currently a local JSON file, not a database.
  • Actual message delivery depends on the host runtime exposing a compatible send interface.
安全使用建议
This skill appears to do what it says: analyze YouTube videos, manage a local watchlist, and optionally send notifications through whatever delivery adapter the host runtime provides. Before installing, consider the following: - Watchlist persistence: it writes a local JSON file (default ./data/watchlist.json or a path you set). Ensure watchlist_path is not pointed at any sensitive system file. - Optional API key: providing a YouTube Data API key (config.youtube_api_key or YOUTUBE_API_KEY env var) enables the official API path; if you don’t provide it the skill will scrape public YouTube pages/feeds which is more brittle but expected. - Notifications: the skill will call host-provided delivery methods (context.deliver, context.notify, channel.send, etc.) to push updates. Review what delivery adapters your OpenClaw runtime exposes so you know where notifications may be sent. - No external installers or obscure network endpoints are present in the package — only youtube/google endpoints and the host runtime delivery interfaces are used. If you want higher assurance before installing: review the omitted truncated functions (public-source resolver) to confirm only YouTube/Google endpoints are contacted, and decide if you are comfortable with a local watchlist file being created/updated by the skill.
功能分析
Type: OpenClaw Skill Name: creator-feed-watch Version: 0.2.0 The creator-feed-watch skill is a well-structured YouTube monitoring and analysis tool. It implements features for normalizing YouTube URLs, fetching video metadata via the YouTube Data API or public RSS feeds, and extracting transcripts for AI-powered summarization. The code uses standard Node.js built-in modules for file persistence (watchlist.json) and follows the OpenClaw pattern for dispatching notifications through host-provided adapters. No evidence of malicious intent, unauthorized data exfiltration, or harmful prompt injection was found in the code or instructions.
能力评估
Purpose & Capability
Name/description match the included code and manifest. The skill normalizes YouTube inputs, fetches transcripts/uploads, summarizes video metadata, and persists a local watchlist — all expected for a YouTube creator tracking/analysis tool. No unrelated cloud credentials or unrelated binaries are requested.
Instruction Scope
SKILL.md describes runtime actions (analyze_video, add_follow, check/notify updates) that map to the code. The instructions limit automated follows (require user confirmation) and document the fallback scraping path when no API key is present. The runtime does read/write the local watchlist JSON and may call host-provided delivery adapters to send notifications; both are within the described scope.
Install Mechanism
No install spec and no external archive downloads are used. Code is included in the package (node runtime) and package.json is minimal. There are no downloads from personal servers or URL shorteners. This is a normal packaged skill.
Credentials
The skill does not require any env vars but supports an optional YouTube Data API key. The code checks context.config.youtube_api_key and falls back to process.env.YOUTUBE_API_KEY; manifest lists config.youtube_api_key as an optional config. This is proportionate for the functionality, but note the code will read process.env.YOUTUBE_API_KEY (not declared as a required env var) if provided.
Persistence & Privilege
The skill persists a local JSON watchlist (default ./data/watchlist.json or a user-configured path). It does not request global 'always' inclusion, nor does it modify other skills' configs. Persisting to a user-configurable path is expected, but verify watchlist_path to avoid overwriting sensitive files.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install creator-feed-watch
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /creator-feed-watch 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v0.2.0
- Added video content analysis: Paste a YouTube video link to get an AI-powered summary, key points, and a "why it matters" section. - Prioritizes transcript extraction for in-depth analysis; falls back to title/description if no transcript is available. - Includes a robust, emoji-guided template for clear mobile-friendly presentation. - Now supports multi-language auto-reply, matching the user's message language. - Prompts users to follow creators after video analysis; following remains optional and based on user reply. - Watchlist and upload tracking features remain — can notify users of new videos from followed creators.
v0.0.0
creator-feed-watch 1.0.0 – Initial Release - Normalize YouTube links, handles, and IDs into structured follow targets. - Fetch and track latest uploads for followed channels. - Generate concise update messages for new YouTube videos. - Maintain a local watchlist of followed creators. - Supports both YouTube API and public data fallback. - Prepares update notifications if delivery channels are available.
元数据
Slug creator-feed-watch
版本 0.2.0
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 2
常见问题

Content Creator Skill 是什么?

Track YouTube creators and analyze video content. Paste a YouTube link to get an AI-powered breakdown (summary, key points, why it matters), then optionally... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 148 次。

如何安装 Content Creator Skill?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install creator-feed-watch」即可一键安装,无需额外配置。

Content Creator Skill 是免费的吗?

是的,Content Creator Skill 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Content Creator Skill 支持哪些平台?

Content Creator Skill 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 Content Creator Skill?

由 Lee Lin(@glin23)开发并维护,当前版本 v0.2.0。

💬 留言讨论