What it does

This is a Claude Code Skill that ingests content from 15+ sources—WeChat articles, paywalled news, YouTube, podcasts, EPUBs, even ZIP dumps—then uploads it to Google NotebookLM and transforms it into whatever format you ask for: podcast audio, slide decks, mind maps, quizzes. You type a URL or file path in natural language; it handles the rest.

The interesting bit

The paywall bypass is the real engineering here. It runs a six-level fallback cascade—proxy services, bot user-agent rotation, AMP extraction, archive.today, Google Cache—before giving up. The techniques are borrowed from the “Bypass Paywalls Clean” browser extension, adapted into an automated pipeline. For Chinese users, there’s also MCP browser simulation for WeChat articles and Get笔记 API integration for transcribing 小宇宙/喜马拉雅 podcasts.

Key highlights

• 300+ paywalled sites supported: NYT, WSJ, FT, Economist, The Information, and regional outlets across US/UK/Germany/France/Australia/China
• Six-tier bypass cascade with automatic strategy selection per site
• Native handling of Chinese platforms: WeChat (via MCP browser), 小宇宙/喜马拉雅/B站 podcasts
• Output formats: podcast MP3, PPT PDF, mind map JSON, quiz, report, infographic, flashcards
• Multi-source mixing: can combine paywalled article + YouTube + EPUB into single synthesized report
• One-command install via install.sh; requires only Python 3.9+ and Git

Caveats

• Requires separate NotebookLM authentication and optional Get笔记 API key for podcast transcription
• Paywall bypass is an arms race; the README doesn’t specify success rates per tier or how often manual intervention is needed
• Some outputs (video, infographic) are mentioned but not demonstrated with concrete examples in the sources

Verdict

Heavy NotebookLM users who consume a lot of paywalled journalism or Chinese social media content will save serious time. If you already have your own scraping setup or don’t use NotebookLM, this is mostly glue code you could skip.