joeseesun/anything-to-notebooklm
Overview
This skill automates collecting content from multiple sources and uploads it to NotebookLM, then optionally generates outputs like podcasts, PPTs, mind maps, quizzes, videos, reports, infographics, or flashcards. It supports WeChat public articles, web pages, YouTube, PDFs, Office files, EPUB/Markdown, images (OCR), audio, CSV/JSON/XML, and ZIP bundles. The tool streamlines conversion, upload, generation, and download into a repeatable CLI-backed workflow.
How this skill works
The skill detects the input type (URL, local path, search keywords, or pasted text), converts non-native formats to text/Markdown via markitdown (or OCR/transcription for images/audio), and creates a NotebookLM notebook and sources using the notebooklm CLI. If a generation intent is provided, it triggers NotebookLM generate commands, waits for artifacts to finish, and downloads the resulting files to a local path. MCP is used for fetching WeChat articles; WebSearch can aggregate search results into a single source.
When to use it
- Turn articles, videos, or local documents into audio summaries (podcasts) for commuting or accessibility
- Create slide decks from research notes, web posts, and PDFs for team presentations
- Produce mind maps or infographics to visualize complex articles or long reports
- Extract and transcribe audio or scanned documents into editable text for analysis
- Bundle mixed sources (webpage + video + PDF) into one NotebookLM project and generate combined outputs
Best practices
- Configure the MCP weixin-reader server and verify notebooklm login before first use
- Keep source files moderate in size; split very long documents to avoid generation failures
- Wait for upload/processing to complete before starting generation to prevent task errors
- Provide clear natural-language generation instructions (style, length, format) to improve output quality
- Respect copyright: use outputs for personal study or internal team use and avoid commercial redistribution
Example use cases
- Convert a WeChat article into a 5–8 minute podcast for commuting listening
- Generate a 15–slide PPT from a research PDF, a related blog post, and a YouTube talk
- Create a mind map from a YouTube lecture by extracting subtitles and structuring key topics
- Search the web for a topic, aggregate top results, upload to NotebookLM, and generate a concise report
- Batch-process a ZIP of mixed documents (DOCX, PDF, images) into a single NotebookLM notebook and export flashcards
FAQ
Install and configure the MCP weixin-reader server in your Claude config and run notebooklm login to authenticate before first use.
What happens if NotebookLM generation fails?
Common causes are very short/very long content, or service issues. Retry after trimming content, or choose a different output type; check notebooklm artifact list for status.
