joeseesun/qiaomu-anything-to-notebooklm

What it solves

This project provides a way to convert diverse content sources—including paywalled news articles, podcasts, and ebooks—into structured AI-generated formats like podcasts, PPTs, mind maps, and quizzes. It eliminates the manual effort of extracting content from difficult-to-scrape sources and uploading them to Google NotebookLM for processing.

How it works

It functions as a Claude Code Skill that uses natural language commands to trigger a pipeline:

1. Content Acquisition: It identifies the source type (e.g., WeChat, X/Twitter, YouTube, PDF) and uses specialized tools to fetch the content. This includes a 6-layer cascading strategy to bypass paywalls on over 300 news sites.
2. Processing: It converts files using tools like Microsoft MarkItDown and transcribes audio via the Get笔记 API.
3. Integration: The processed content is uploaded to Google NotebookLM via an API.
4. Generation: It triggers NotebookLM to generate the target output (e.g., an MP3 podcast or a PDF slide deck).

Who it’s for

Researchers, students, and content creators who need to quickly synthesize information from multiple disparate sources and transform them into digestible formats for study or presentation.

Highlights

• Extensive Source Support: Handles 15+ formats including WeChat, X/Twitter, YouTube, EPUB, and Office documents.
• Paywall Bypassing: Employs a sophisticated 6-level strategy (including Googlebot UA spoofing and archive.today) to access restricted news content.
• Deep Analysis Mode: A recursive three-round questioning strategy (Overview $\rightarrow$ Deep Dive $\rightarrow$ Synthesis) to extract high-level insights from documents.
• Automated Workflow: Fully automates the path from a URL or file path to a final generated asset (e.g., .mp3, .pdf, .json).