Agent skill

Multi-Source to NotebookLM Processor

by joeseesun in joeseesun/qiaomu-anything-to-notebooklm

Collects content from WeChat articles, web pages, YouTube, podcasts, documents and more, uploads it to NotebookLM and generates podcasts, slides or mind maps.

MITAuto-check passedKnowledge Management

SKILL.md written in Chinese; this summary is our English description.

Install Multi-Source to NotebookLM Processor

skills CLI
$ npx skills add joeseesun/qiaomu-anything-to-notebooklm --skill qiaomu-anything-to-notebooklm -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install joeseesun/qiaomu-anything-to-notebooklm qiaomu-anything-to-notebooklm --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
qiaomu-anything-to-notebooklm
GitHub stars
6.2k
Token cost
~3.6k tokens
SKILL.md length
1,008 words
Files
21 (incl. scripts)
Skills in repo
1
Repo updated
First seen
Licence
MIT

At a glance

Collects content from WeChat articles, web pages, YouTube, podcasts, documents and more, uploads it to NotebookLM and generates podcasts, slides or mind maps.

  • Works in 12 steps: 微信公众号文章 → 任意网页链接 → 播客(小宇宙/喜马拉雅)+ B站视频 → …
  • Turning a WeChat article or web page into a NotebookLM podcast
  • SKILL.md covers 支持的内容源, 前置条件 and 触发方式
  • Runs Python and Shell scripts from its folder; calls python, pip and playwright; reaches mp.weixin.qq.com and youtube.com; needs GETNOTE_API_KEY

What it does

Written in Chinese, the skill fetches content from many sources, uploads it to NotebookLM and generates outputs such as podcasts, PPT slides and mind maps from natural-language instructions. Sources include WeChat official account articles through an MCP server, any public web page, podcasts and Bilibili videos transcribed through the Get笔记 API, X/Twitter posts fetched through a proxy cascade, Word, PowerPoint and Excel files, PDFs and EPUB books, images and scans with OCR, audio, CSV, JSON, XML, ZIP archives, plain text and search keywords.

YouTube links are passed straight to NotebookLM, which reads subtitles and metadata itself, and the skill forbids using yt-dlp or browser automation for them. It also includes a fallback for paywalled news sites that tries alternate user agents, referers, AMP pages and web archives. Prerequisites are installing and configuring the bundled WeChat MCP server, restarting Claude Code and running notebooklm login before first use. A deep analysis mode and automatic Feishu document creation are also supported.

When your agent uses it

  • Turning a WeChat article or web page into a NotebookLM podcast
  • Making a PPT or mind map out of a YouTube video, PDF or EPUB
  • Transcribing a Xiaoyuzhou or Ximalaya podcast episode and analyzing it
  • Processing a mix of file types and links into one NotebookLM notebook

Example prompts

  • “Turn this WeChat article into a podcast in NotebookLM.”
  • “Make a slide deck from this YouTube video using NotebookLM.”
  • “Upload ./book.epub to NotebookLM and generate a mind map.”
  • “Transcribe this Ximalaya episode and do a deep analysis.”

Requirements

  • The notebooklm CLI, signed in with notebooklm login
  • The bundled WeChat MCP server, for WeChat articles
  • Access to the Get笔记 API, for podcast and Bilibili transcripts

Workflow steps

12 steps, taken from the step headings in SKILL.md.

  1. 微信公众号文章
  2. 任意网页链接
  3. 播客(小宇宙/喜马拉雅)+ B站视频
  4. X/Twitter 帖子
  5. 付费墙网站自动绕过
  6. YouTube 视频
  7. Office 文档
  8. 播客/音频平台
  9. 电子书与文档
  10. 图片与扫描件
  11. 音频文件
  12. 结构化数据

What it can do on your machine

Read from SKILL.md and the folder at commit cea6cee. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Ships 1 file in scripts/ (Python and Shell, from the files we listed), which the agent can run.

    Shell commands in SKILL.md call:

    • python
    • pip
    • playwright
    • python3
    • bash
    • markitdown

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    Hosts in commands or code, which the agent is likely to contact:

    • mp.weixin.qq.com
    • youtube.com
    • youtu.be
    • x.com

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names these keys or tokens, usually read from environment variables:

    • GETNOTE_API_KEY

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Multi-Source to NotebookLM Processor loads about 3.6k tokens when it runs. Until then it costs about 35 tokens; SKILL.md has 1,008 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~35
When it runs · the whole SKILL.md, loaded when a task matches
~3.6k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.

SKILL.md

The full file from joeseesun/qiaomu-anything-to-notebooklm at commit cea6cee, republished under its MIT licence (© joeseesun). 1,008 words, ~3,572 tokens.

Download SKILL.mdSave it as .claude/skills/qiaomu-anything-to-notebooklm/SKILL.md (or your agent's skills folder). This skill also uses 20 other files; get the full folder from GitHub.
name
qiaomu-anything-to-notebooklm
description
多源内容智能处理器:支持微信公众号、网页、YouTube、播客(小宇宙/喜马拉雅)、PDF、Markdown等,自动上传到NotebookLM并生成播客/PPT/思维导图等多种格式。支持深度分析模式和飞书文档自动创建
user-invocable
true
homepage
https://github.com/joeseesun/qiaomu-anything-to-notebooklm

多源内容 → NotebookLM 智能处理器

自动从多种来源获取内容,上传到 NotebookLM,并根据自然语言指令生成播客、PPT、思维导图等多种格式。

支持的内容源

1. 微信公众号文章

通过 MCP 服务器自动抓取微信公众号文章内容(绕过反爬虫)

2. 任意网页链接

支持任何公开可访问的网页(新闻、博客、文档等)

3. 播客(小宇宙/喜马拉雅)+ B站视频

通过 Get笔记 API 获取完整转写文本(带时间戳),支持小宇宙、喜马拉雅、B站视频等音频/视频平台

4. X/Twitter 帖子

通过内置代理级联(r.jina.ai → defuddle.md → agent-fetch)抓取推文内容(含长推文线程),转为 Markdown

5. 付费墙网站自动绕过

自动检测并绕过 NYT、WSJ、FT、Economist、Bloomberg、Medium 等 300+ 付费网站的付费墙。策略:UA 伪装(Googlebot/Bingbot)→ Referer 伪装(Google/Facebook)→ AMP 页面 → archive.today 存档

5. YouTube 视频

直接传递给 NotebookLM! NotebookLM 原生支持 YouTube 链接,会自动提取视频字幕和元数据,无需手动下载字幕或转写。禁止使用 yt-dlp 或浏览器自动化提取字幕。

4. Office 文档
  • Word (DOCX) - 保留表格和格式
  • PowerPoint (PPTX) - 提取幻灯片和备注
  • Excel (XLSX) - 表格数据
5. 播客/音频平台
  • 小宇宙 (xiaoyuzhoufm.com) - 通过 Get笔记 API 获取完整转写
  • 喜马拉雅 (ximalaya.com) - 通过 Get笔记 API 获取完整转写
  • B站视频 (bilibili.com) - 通过 Get笔记 API 获取完整转写
  • 其他音频链接 - 通过 Get笔记 API 获取转写
5. 电子书与文档
  • PDF - 全文提取
  • EPUB - 电子书全文提取
  • Markdown (.md) - 原生支持
6. 图片与扫描件
  • Images (JPEG, PNG, GIF, WebP) - OCR 识别文字
  • 扫描的 PDF 文档 - OCR 提取文字
7. 音频文件
  • Audio (WAV, MP3) - 语音转文字
8. 结构化数据
  • CSV - 逗号分隔数据
  • JSON - JSON 数据
  • XML - XML 文档
9. 压缩包
  • ZIP - 自动解压并处理所有支持的文件
10. 纯文本

直接输入或粘贴的文本内容

11. 搜索关键词

通过 Web Search 搜索关键词,汇总多个来源的信息

前置条件

1. 安装 wexin-read-mcp

MCP 服务器已安装在:~/.claude/skills/qiaomu-anything-to-notebooklm/wexin-read-mcp/

配置 MCP(需要手动添加到 Claude 配置文件):

macOS: 编辑 ~/.claude/config.json

json
{
  "primaryApiKey": "any",
  "mcpServers": {
    "weixin-reader": {
      "command": "python",
      "args": [
        "/Users/joe/.claude/skills/qiaomu-anything-to-notebooklm/wexin-read-mcp/src/server.py"
      ]
    }
  }
}

配置后需要重启 Claude Code。

2. notebooklm 认证

首次使用前必须认证:

bash
notebooklm login
notebooklm list  # 验证认证成功

触发方式

微信公众号文章
  • /qiaomu-anything-to-notebooklm [微信文章链接]
  • "把这篇微信文章传到NotebookLM"
  • "把这篇微信文章生成播客"
网页链接
  • "把这个网页做成播客 [URL]"
  • "这篇文章帮我做成PPT [URL]"
  • "帮我分析这个网页 [URL]"
播客(小宇宙/喜马拉雅/B站)
  • "把这个播客生成播客 [小宇宙链接]"
  • "这个小宇宙节目帮我做成PPT [链接]"
  • "深度解读这期播客 [喜马拉雅链接]"
  • "把这个B站视频转写传到NotebookLM [bilibili链接]"
  • "B站视频帮我生成思维导图 [bilibili链接]"
X/Twitter 帖子
  • "把这条推文传到NotebookLM [x.com链接]"
  • "这篇推文线程帮我生成报告 [x.com链接]"
  • "深度分析这条推文 [twitter.com链接]"
YouTube 视频
  • 直接传 URL 给 NotebookLM,不下载字幕! NotebookLM 原生支持 YouTube
  • "把这个YouTube视频做成播客 [YouTube URL]"
  • "这个视频帮我生成思维导图 [YouTube URL]"
本地文件
  • "把这个PDF上传到NotebookLM /path/to/file.pdf"
  • "这个Markdown文件生成PPT /path/to/file.md"
  • "这个EPUB电子书生成播客 /path/to/book.epub"
  • "把这个Word文档做成思维导图 /path/to/doc.docx"
  • "这个PowerPoint生成Quiz /path/to/slides.pptx"
  • "把这个扫描PDF做成报告 /path/to/scan.pdf"(自动OCR)
搜索关键词
  • "搜索 'AI发展趋势' 并生成报告"
  • "搜索关于'量子计算'的资料做成播客"
混合使用
  • "把这篇文章、这个视频和这个PDF一起上传,生成一份报告"
深度分析模式(递归提问)
  • "深度分析这本书 /path/to/book.epub"
  • "提炼这篇文章的核心观点 [URL]"
  • "递归提问分析这个PDF /path/to/file.pdf"
  • "帮我深度解读这个视频 [YouTube URL]"
深度分析 + 飞书文档
  • "深度分析这本书并写入飞书 /path/to/book.epub"
  • "分析这篇文章后创建飞书文档 [URL]"
  • "递归提问并生成飞书文档 /path/to/file.pdf"

自然语言 → NotebookLM 功能映射

用户说的话识别意图NotebookLM 命令
"生成播客" / "做成音频" / "转成语音"audiogenerate audio
"做成PPT" / "生成幻灯片" / "做个演示"slide-deckgenerate slide-deck
"画个思维导图" / "生成脑图" / "做个导图"mind-mapgenerate mind-map
"生成Quiz" / "出题" / "做个测验"quizgenerate quiz
"做个视频" / "生成视频"videogenerate video
"生成报告" / "写个总结" / "整理成文档"reportgenerate report
"做个信息图" / "可视化"infographicgenerate infographic
"生成数据表" / "做个表格"data-tablegenerate data-table
"做成闪卡" / "生成记忆卡片"flashcardsgenerate flashcards
"深度分析" / "提炼核心观点" / "递归提问" / "深度解读"deep-analysis自动生成10个问题并递归提问
"写入飞书" / "创建飞书文档" / "生成飞书文档" / "保存到飞书"feishu创建飞书文档并写入内容

如果没有明确指令,默认只上传不生成任何内容,等待用户后续指令。

工作流程

Step 1: 识别内容源类型

Claude 自动识别输入类型:

输入特征识别为处理方式
https://mp.weixin.qq.com/s/微信公众号MCP 工具抓取
https://youtube.com/... 或 https://youtu.be/...YouTube直接传递给 NotebookLM
xiaoyuzhoufm.com 或 ximalaya.com 或 bilibili.com播客/视频Get笔记 API 转写 → TXT
x.com 或 twitter.comX/Twitter 帖子内置代理级联抓取 → TXT
https:// 或 http://(付费网站)付费墙网页内置付费墙绕过(UA伪装+archive.today)→ TXT
https:// 或 http://网页直接传递给 NotebookLM
/path/to/file.pdfPDF 文件markitdown 转 Markdown → TXT
/path/to/file.epubEPUB 电子书Python ebooklib 提取文本 → TXT(避免 Calibre)
/path/to/file.docxWord 文档markitdown 转 Markdown → TXT
/path/to/file.pptxPowerPointmarkitdown 转 Markdown → TXT
/path/to/file.xlsxExcelmarkitdown 转 Markdown → TXT
/path/to/file.mdMarkdown直接上传
/path/to/image.jpg图片(OCR)markitdown OCR → TXT
/path/to/audio.mp3音频markitdown 转录 → TXT
/path/to/file.zipZIP 压缩包解压 → markitdown 批量转换
关键词(无URL,无路径)搜索查询WebSearch → 汇总 → TXT
Step 2: 获取内容

微信公众号:

  • 使用 MCP 工具 read_weixin_article
  • 返回:title, author, publish_time, content
  • 保存为 TXT:/tmp/weixin_{title}_{timestamp}.txt

播客/视频(小宇宙/喜马拉雅/B站):

  • 通过 Get笔记 API 获取完整转写文本
  • 调用 python3 ~/.claude/skills/qiaomu-anything-to-notebooklm/scripts/get_podcast_transcript.py <URL>
  • 脚本自动执行:创建链接笔记 → 等待转写 → 获取全文 → 保存 TXT
  • 返回 TXT 路径和标题
  • 依赖:Get笔记 API Key(环境变量 GETNOTE_API_KEY、GETNOTE_CLIENT_ID)+ Web Token(~/.claude/skills/getnote/tokens.json)

X/Twitter 帖子:

  • 通过内置代理级联抓取推文内容(r.jina.ai → defuddle.md → agent-fetch)
  • 调用 bash ~/.claude/skills/qiaomu-anything-to-notebooklm/scripts/fetch_url.sh "https://x.com/..." 获取 Markdown 内容
  • 自动处理 X 登录墙和错误页面
  • 保存为 TXT 后上传到 NotebookLM

网页:

  • 直接使用 notebooklm source add <URL>
  • NotebookLM 自动提取内容
  • 付费墙绕过:遇到付费网站时,fetch_url.sh 自动启用多重绕过策略

YouTube 🔴 特殊规则(最重要!):

  • 直接传递 URL 给 NotebookLM! notebooklm source add <YouTube_URL>
  • 禁止使用 yt-dlp、yt-search-download、whisper、浏览器自动化等手段下载字幕
  • NotebookLM 原生支持 YouTube,会自动提取字幕和元数据
  • 这是最快速、最高效的方式,不需要任何中间步骤
    1. r.jina.ai — 通常能绕过软付费墙
    2. Googlebot/Bingbot UA 伪装 — 模拟搜索引擎爬虫(网站为了 SEO 通常给爬虫全文)
    3. Referer 伪装 — 伪装来自 Google/Facebook(社交引流豁免)
    4. AMP 页面 — AMP 版本通常没有付费墙
    5. archive.today — 从网页存档获取全文
    • 支持的付费网站:NYT、WSJ、FT、Economist、Bloomberg、Washington Post、New Yorker、Wired、The Atlantic、Medium、MIT Technology Review、SCMP 等 300+ 站点

Office 文档/电子书/PDF:

  • EPUB:使用 Python ebooklib + BeautifulSoup 直接提取文本(避免 Calibre 架构问题)
  • 其他格式:使用 markitdown 转换为 Markdown
  • 命令:markitdown /path/to/file.docx -o /tmp/converted.md
  • 保存为 TXT:/tmp/{filename}_converted_{timestamp}.txt

本地 Markdown:

  • 直接上传:notebooklm source add /path/to/file.md

图片(OCR):

  • markitdown 自动 OCR 识别文字
  • 提取 EXIF 元数据
  • 保存为 TXT

音频文件:

  • markitdown 自动转录语音为文字
  • 提取音频元数据
  • 保存为 TXT

ZIP 压缩包:

  • 自动解压到临时目录
  • 遍历所有支持的文件
  • 批量使用 markitdown 转换
  • 合并为单个 TXT 或多个 Source

搜索关键词:

  • 使用 WebSearch 工具搜索关键词
  • 汇总前 3-5 条结果
  • 保存为 TXT:/tmp/search_{keyword}_{timestamp}.txt
Step 3: 上传到 NotebookLM

调用 notebooklm skill:

bash
notebooklm create "{title}"  # 创建新笔记本
notebooklm source add /tmp/weixin_xxx.txt --title "{title}"  # 上传文件

注意:NotebookLM 会自动处理上传的文件,无需手动等待。

Show full SKILL.md (401 more words)Show less
Step 4: 深度分析模式(可选)

如果用户指定了"深度分析"、"递归提问"等意图,自动执行:

bash
# 仅深度分析
python ~/.claude/skills/qiaomu-anything-to-notebooklm/main.py \
  /path/to/file.epub --deep-analysis

# 深度分析 + 自动创建飞书文档
python ~/.claude/skills/qiaomu-anything-to-notebooklm/main.py \
  /path/to/file.epub --deep-analysis --to-feishu

深度分析流程:

  1. 上传内容到 NotebookLM
  2. 根据内容类型自动生成 10 个深度问题
  3. 依次向 NotebookLM 提问并收集答案
  4. 返回结构化 JSON 数据(包含问题、答案、统计信息)
  5. (可选)如果指定 --to-feishu,自动创建飞书文档并写入问答内容

问题类型:

  • 书籍/文档:核心观点、金句、论证逻辑、实践建议、局限性等
  • 视频:目标受众、关键数据、叙事结构、精华版内容等
  • 文章/网页:写作目的、数据支撑、作者立场、个人启发等

输出格式:

json
{
  "status": "success",
  "title": "书名/标题",
  "content_type": "epub/document/url",
  "questions": ["问题1", "问题2", ...],
  "answers": ["答案1", "答案2", ...],
  "total_questions": 10,
  "answered": 10
}
Step 5: 根据意图生成内容(可选)

如果用户指定了处理意图,自动调用对应命令:

意图命令等待下载
audionotebooklm generate audioartifact waitdownload audio ./output.mp3
slide-decknotebooklm generate slide-deckartifact waitdownload slide-deck ./output.pdf
mind-mapnotebooklm generate mind-mapartifact waitdownload mind-map ./map.json
quiznotebooklm generate quizartifact waitdownload quiz ./quiz.md --format markdown
videonotebooklm generate videoartifact waitdownload video ./output.mp4
reportnotebooklm generate reportartifact waitdownload report ./report.md
infographicnotebooklm generate infographicartifact waitdownload infographic ./infographic.png
flashcardsnotebooklm generate flashcardsartifact waitdownload flashcards ./cards.md --format markdown

生成流程:

  1. 发起生成请求(返回 task_id)
  2. 等待生成完成(artifact wait <task_id>)
  3. 下载生成的文件到本地
  4. 告知用户文件路径

完整示例

示例 1:微信公众号文章 → 播客

用户输入:

把这篇文章生成播客 https://mp.weixin.qq.com/s/abc123xyz

执行流程:

  1. 识别为微信公众号链接
  2. MCP 工具抓取文章内容
  3. 创建 TXT 文件
  4. 上传到 NotebookLM
  5. 生成播客(generate audio)
  6. 下载播客到本地

输出:

✅ 微信文章已转换为播客!

📄 文章:深度学习的未来趋势
👤 作者:张三
📅 发布:2026-01-20

🎙️ 播客已生成:
📁 文件:/tmp/weixin_深度学习的未来趋势_podcast.mp3
⏱️ 时长:约 8 分钟
📊 大小:12.3 MB
示例 2:YouTube 视频 → 思维导图

用户输入:

这个视频帮我画个思维导图 https://www.youtube.com/watch?v=abc123

执行流程:

  1. 识别为 YouTube 链接
  2. 直接传递给 NotebookLM(自动提取字幕)
  3. 生成思维导图(generate mind-map)
  4. 下载思维导图

输出:

✅ YouTube 视频已转换为思维导图!

🎬 视频:Understanding Quantum Computing
⏱️ 时长:23 分钟

🗺️ 思维导图已生成:
📁 文件:/tmp/youtube_quantum_computing_mindmap.json
📊 节点数:45 个
示例 3:搜索关键词 → 报告

用户输入:

搜索 'AI发展趋势 2026' 并生成报告

执行流程:

  1. 识别为搜索查询
  2. WebSearch 搜索关键词
  3. 汇总前 5 条结果
  4. 创建 TXT 文件
  5. 上传到 NotebookLM
  6. 生成报告(generate report)

输出:

✅ 搜索结果已生成报告!

🔍 关键词:AI发展趋势 2026
📊 来源:5 篇文章

📄 报告已生成:
📁 文件:/tmp/search_AI发展趋势2026_report.md
📝 章节:7 个
📊 大小:15.2 KB
示例 4:混合多源 → PPT

用户输入:

把这篇文章、这个视频和这个PDF一起做成PPT:
- https://example.com/article
- https://youtube.com/watch?v=xyz
- /Users/joe/Documents/research.pdf

执行流程:

  1. 创建新 Notebook
  2. 依次添加 3 个 Source
  3. 基于所有 Source 生成 PPT

输出:

✅ 多源内容已整合为PPT!

📚 内容源:
  1. 网页文章:AI in 2026
  2. YouTube:Future of AI
  3. PDF:Research Notes (12 页)

📊 PPT 已生成:
📁 文件:/tmp/multi_source_slides.pdf
📄 页数:25 页
📦 大小:3.8 MB
示例 5: EPUB 电子书 → 播客

用户输入:

把这本电子书做成播客 /Users/joe/Books/sapiens.epub

执行流程:

  1. 识别为 EPUB 文件
  2. markitdown 转换为 Markdown
  3. 保存为 TXT
  4. 上传到 NotebookLM
  5. 生成播客

输出:

✅ EPUB 电子书已转换为播客!

📚 电子书:Sapiens: A Brief History of Humankind
📄 页数:约 450 页
📊 字数:约 15 万字

🎙️ 播客已生成:
📁 文件:/tmp/sapiens_podcast.mp3
⏱️ 时长:约 45 分钟(精华版)
📊 大小:48.2 MB
示例 6:Word 文档 → Quiz

用户输入:

这个Markdown生成Quiz /Users/joe/notes/machine_learning.md

执行流程:

  1. 识别为本地 Markdown 文件
  2. 直接上传到 NotebookLM
  3. 生成 Quiz(generate quiz)

输出:

✅ Markdown 已转换为Quiz!

📄 文件:machine_learning.md
📊 大小:8.5 KB

📝 Quiz 已生成:
📁 文件:/tmp/machine_learning_quiz.md
❓ 题目:15 道(10选择 + 5简答)

错误处理

URL 格式错误
❌ 错误:URL 格式不正确

必须是微信公众号文章链接:
https://mp.weixin.qq.com/s/xxx

你提供的链接:https://example.com
文章获取失败
❌ 错误:无法获取文章内容

可能原因:
1. 文章已被删除
2. 文章需要登录查看(暂不支持)
3. 网络连接问题
4. 微信反爬虫拦截(请稍后重试)

建议:
- 检查链接是否正确
- 等待 2-3 秒后重试
- 或手动复制文章内容
NotebookLM 认证失败
❌ 错误:NotebookLM 认证失败

请运行以下命令重新登录:
  notebooklm login

然后验证:
  notebooklm list
生成任务失败
❌ 错误:播客生成失败

可能原因:
1. 文章内容太短(< 100 字)
2. 文章内容太长(> 50万字)
3. NotebookLM 服务异常

建议:
- 检查文章长度是否适中
- 稍后重试
- 或尝试其他格式(如生成报告)

高级功能

1. 多意图处理

用户可以一次性指定多个处理任务:

这篇文章帮我生成播客和PPT https://mp.weixin.qq.com/s/abc123

Skill 会依次执行:

  1. 生成播客
  2. 生成 PPT
2. 自定义 Notebook

默认每篇文章创建新 Notebook,也可以指定已有 Notebook:

把这篇文章加到我的【AI研究】笔记本 https://mp.weixin.qq.com/s/abc123

Skill 会:

  1. 搜索名为"AI研究"的 Notebook
  2. 将文章添加为新 Source
  3. 基于所有 Sources 生成内容
3. 自定义生成指令

为生成任务添加具体要求:

这篇文章生成播客,要求:轻松幽默的风格,时长控制在5分钟

Skill 会将要求作为 instructions 传给 NotebookLM。

注意事项

  1. 频率限制:

    • 每次请求间隔 > 2 秒,避免被微信封禁
    • NotebookLM 生成任务有并发限制(最多 3 个同时进行)
  2. 内容长度:

    • 微信文章通常 1000-5000 字,适合生成播客(3-8 分钟)
    • 超过 10000 字的长文可能需要更长生成时间
    • 少于 500 字的短文可能生成效果不佳
  3. 版权遵守:

    • 仅用于个人学习研究
    • 遵守微信公众号的版权规定
    • 生成的内容不得用于商业用途
  4. 生成时间:

    • 播客:2-5 分钟
    • 视频:3-8 分钟
    • PPT:1-3 分钟
    • 思维导图:1-2 分钟
    • Quiz/闪卡:1-2 分钟
  5. 文件清理:

    • TXT 源文件保存在 /tmp/,系统重启后自动清理
    • 生成的文件(MP3/PDF/MD 等)默认保存在 /tmp/
    • 可以指定自定义保存路径

相关 Skills

  • notebooklm - NotebookLM 核心功能
  • notebooklm-deep-analyzer - 深度分析 NotebookLM 内容
  • markitdown - 转换其他格式文档

配置 MCP(重要)

⚠️ 第一次使用前必须配置

编辑 ~/.claude/config.json:

json
{
  "primaryApiKey": "any",
  "mcpServers": {
    "weixin-reader": {
      "command": "python",
      "args": [
        "/Users/joe/.claude/skills/qiaomu-anything-to-notebooklm/wexin-read-mcp/src/server.py"
      ]
    }
  }
}

配置后重启 Claude Code!

故障排查

1. MCP 工具未找到
bash
# 测试 MCP 服务器
python ~/.claude/skills/qiaomu-anything-to-notebooklm/wexin-read-mcp/src/server.py

# 如果报错,检查依赖
cd ~/.claude/skills/qiaomu-anything-to-notebooklm/wexin-read-mcp
pip install -r requirements.txt
playwright install chromium
2. NotebookLM 命令失败
bash
# 检查认证状态
notebooklm status

# 重新登录
notebooklm login

# 验证
notebooklm list
3. 文件权限问题
bash
# 确保临时目录可写
chmod 755 /tmp

# 测试写入
touch /tmp/test.txt && rm /tmp/test.txt
4. 生成任务卡住
bash
# 检查任务状态
notebooklm artifact list

# 如果显示 "pending" 超过 10 分钟,取消重试
# (目前 CLI 不支持取消,需要在网页端操作)

典型使用场景

场景 1:快速学习
我想学习这篇文章,帮我生成播客,上下班路上听
链接:https://mp.weixin.qq.com/s/abc123

→ 生成 8 分钟播客,通勤时间听完

场景 2:分享给团队
这篇文章不错,做成PPT分享给团队
https://mp.weixin.qq.com/s/abc123

→ 生成 15 页 PPT,直接用于团队分享

场景 3:复习巩固
这篇技术文章帮我出题,想测试一下掌握程度
https://mp.weixin.qq.com/s/abc123

→ 生成 10 道选择题 + 5 道简答题

场景 4:可视化理解
这篇文章概念比较多,画个思维导图帮我理清结构
https://mp.weixin.qq.com/s/abc123

→ 生成思维导图,一目了然


Skill 创建时间:2026-01-25 最后更新:2026-01-25 版本:v1.0.0

© joeseesun, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 20 other files (scripts) in the repository root of joeseesun/qiaomu-anything-to-notebooklm.

  • SKILL.md
  • .gitignore
  • LICENSE
  • README.md
  • check_env.py
  • feishu-read-mcp/README.md
  • feishu-read-mcp/install.sh
  • feishu-read-mcp/requirements.txt
  • feishu-read-mcp/src/__init__.py
  • feishu-read-mcp/src/image_handler.py
  • feishu-read-mcp/src/parser.py
  • feishu-read-mcp/src/scraper.py
  • feishu-read-mcp/src/server.py
  • feishu-read-mcp/test.py
  • install.sh
  • main.py
  • package.sh
  • requirements.txt
  • scripts
  • … and 2 more

Open the folder on GitHubat commit cea6cee

Compare with similar skills

Multi-Source to NotebookLM Processor next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Multi-Source to NotebookLM Processor compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Multi-Source to NotebookLM Processor this skilljoeseesun/qiaomu-anything-to-notebooklm6.2k—~3.6kAutomated safety check: PassMIT
Nlm Skilliusztinpaul/ai-research-os-workshop1791 repos~6.9kAutomated safety check: PassMIT
NotebookLM CLI Guidejacob-bd/notebooklm-cli256—~3.4kAutomated safety check: WarnMIT
Video To Notelike-attract/video-to-note124—~1kAutomated safety check: PassMIT
NotebooklmMathews-Tom/armory329—~4kAutomated safety check: PassMIT
NotebookLM Research Workflowclaude-world/notebooklm-skill467—~1.8kAutomated safety check: PassMIT

Similar skills

  • Nlm Skill

    iusztinpaul/ai-research-os-workshop

    Expert guide for the NotebookLM CLI (nlm) and MCP server - interfaces for Google NotebookLM.

    179 GitHub starsUsed in 1 repo~6.9k tokens
    Knowledge ManagementAuto-check passed
  • NotebookLM CLI Guide

    jacob-bd/notebooklm-cli

    Guides use of the nlm command-line tool to automate Google NotebookLM: notebooks, sources, research, one-shot questions and generated podcasts, reports, quizzes and slides.

    256 GitHub stars~3.4k tokensUpdated 4 mo ago
    Knowledge ManagementAuto-check: warnings
  • Video To Note

    like-attract/video-to-note

    Generate structured, timestamped Markdown notes from videos (Bilibili, Douyin, YouTube, or local media files) using the local VideoToNo service.

    124 GitHub stars~1k tokensUpdated 7 days ago
    Knowledge ManagementAuto-check passed
  • Notebooklm

    Mathews-Tom/armory

    Full NotebookLM API via notebooklm-py CLI: create notebooks, add sources, generate podcasts, videos, infographics, slides, quizzes, flashcards, mind maps.

    329 GitHub stars~4k tokensUpdated 5 days ago
    Knowledge ManagementAuto-check passed
  • NotebookLM Research Workflow

    claude-world/notebooklm-skill

    Creates NotebookLM notebooks from URLs, text and files, asks cited questions, runs web research and generates audio, slides, quizzes and other artifacts.

    467 GitHub stars~1.8k tokensUpdated 2 mo ago
    Knowledge ManagementAuto-check passed
  • Open Notebook

    agent-skills-hub/agent-skills-hub

    Drives a self-hosted Open Notebook instance to organize sources into notebooks, chat with documents, generate notes and multi-speaker podcasts, and search across material.

    112 GitHub starsUsed in 3 repos~2.6k tokens
    Knowledge ManagementAuto-check passed

Questions about Multi-Source to NotebookLM Processor

What does Multi-Source to NotebookLM Processor do?

Collects content from WeChat articles, web pages, YouTube, podcasts, documents and more, uploads it to NotebookLM and generates podcasts, slides or mind maps. Written in Chinese, the skill fetches content from many sources, uploads it to NotebookLM and generates outputs such as podcasts, PPT slides and mind maps from natural-language instructions. Sources include WeChat official account articles through an MCP server, any public web page, podcasts and Bilibili videos transcribed through the Get笔记 API, X/Twitter posts fetched through a proxy cascade, Word, PowerPoint and Excel files, PDFs and EPUB books, images and scans with OCR, audio, CSV, JSON, XML, ZIP archives, plain text and search keywords.

When should I use Multi-Source to NotebookLM Processor?

Multi-Source to NotebookLM Processor fits situations like: turning a WeChat article or web page into a NotebookLM podcast; making a PPT or mind map out of a YouTube video, PDF or EPUB; transcribing a Xiaoyuzhou or Ximalaya podcast episode and analyzing it; processing a mix of file types and links into one NotebookLM notebook.

How do I install Multi-Source to NotebookLM Processor in Claude Code?

Run `npx skills add joeseesun/qiaomu-anything-to-notebooklm --skill qiaomu-anything-to-notebooklm -a claude-code`. Or copy the skill folder (the joeseesun/qiaomu-anything-to-notebooklm repository) into .claude/skills/qiaomu-anything-to-notebooklm in your project. Claude Code loads it when a task matches its description.

How do I install Multi-Source to NotebookLM Processor in Codex?

Run `npx skills add joeseesun/qiaomu-anything-to-notebooklm --skill qiaomu-anything-to-notebooklm -a codex`. Or copy the skill folder (the joeseesun/qiaomu-anything-to-notebooklm repository) into .agents/skills/qiaomu-anything-to-notebooklm in your project. Codex loads it when a task matches its description.

Can I use Multi-Source to NotebookLM Processor in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add joeseesun/qiaomu-anything-to-notebooklm --skill qiaomu-anything-to-notebooklm -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/qiaomu-anything-to-notebooklm, .gemini/skills/qiaomu-anything-to-notebooklm, .github/skills/qiaomu-anything-to-notebooklm and .opencode/skills/qiaomu-anything-to-notebooklm in your project.

What does Multi-Source to NotebookLM Processor need to run?

Going by SKILL.md and its folder, Multi-Source to NotebookLM Processor needs Python and a shell for the scripts in its folder, the command-line tools its instructions call (python, pip, playwright, python3, bash and markitdown) and credentials named GETNOTE_API_KEY. Our summary lists: The notebooklm CLI, signed in with notebooklm login; The bundled WeChat MCP server, for WeChat articles; Access to the Get笔记 API, for podcast and Bilibili transcripts.

Does Multi-Source to NotebookLM Processor access the network?

SKILL.md names 4 domains. In commands or code: mp.weixin.qq.com, youtube.com, youtu.be and x.com; the agent is likely to contact these when it follows the instructions. This is read from the text; nothing was executed.

Is Multi-Source to NotebookLM Processor safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.

What licence does Multi-Source to NotebookLM Processor use?

Multi-Source to NotebookLM Processor is published under the MIT licence (from the LICENSE file in the skill folder). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Multi-Source to NotebookLM Processor use?

About 3.6k tokens (SKILL.md is roughly 14k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Multi-Source to NotebookLM Processor?

Skills that share tags, products or a category with Multi-Source to NotebookLM Processor: Nlm Skill (iusztinpaul/ai-research-os-workshop, 179 stars), NotebookLM CLI Guide (jacob-bd/notebooklm-cli, 256 stars), Video To Note (like-attract/video-to-note, 124 stars) and Notebooklm (Mathews-Tom/armory, 329 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Multi-Source to NotebookLM Processor?

joeseesun (a GitHub user) maintains it in joeseesun/qiaomu-anything-to-notebooklm, which has 6,204 GitHub stars. The repository was last updated on October 5, 2026.

Source: joeseesun/qiaomu-anything-to-notebooklm on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.