A Claude Code plugin that automates research paper processing with Google's NotebookLM. Fetch papers from Hugging Face Daily Papers, arXiv, or any PDF URL, then automatically create NotebookLM notebooks with AI-generated slide decks and audio overviews.
- Daily Papers Pipeline: Automatically process top papers from Hugging Face Daily Papers
- Single Paper Processing: Quick processing of individual arXiv papers
- Batch Processing: Process multiple papers from a file
- arXiv Search: Search and filter papers by topic, author, or category
- Audio Overviews: Generate podcast-style discussions of papers
- Slide Generation: Create professional slide decks from paper content
- Claude Code v1.0.33 or later
- Claude in Chrome MCP extension installed and connected
- Google account with NotebookLM access
- Active browser session with NotebookLM
claude --plugin-dir /path/to/notebooklm-papers-plugin/plugin install notebooklm-papers
Fetch and process top papers from Hugging Face Daily Papers.
# Process today's top 5 papers
/notebooklm-papers:daily-papers
# Process papers from specific date
/notebooklm-papers:daily-papers 2026-01-08
# Process 10 papers with audio overviews
/notebooklm-papers:daily-papers 2026-01-08 10 --audio
# Use presenter-style slides
/notebooklm-papers:daily-papers --presenterProcess a single arXiv paper.
# By arXiv ID
/notebooklm-papers:process-paper 2601.02151
# By URL
/notebooklm-papers:process-paper https://arxiv.org/abs/2601.02151
# With audio overview
/notebooklm-papers:process-paper 2601.02151 --audio
# Custom notebook name
/notebooklm-papers:process-paper 2601.02151 --name "My Research Paper"Fetch papers without processing (just display information).
# Today's papers
/notebooklm-papers:fetch-papers
# Specific date, more papers
/notebooklm-papers:fetch-papers 2026-01-08 20
# Save to file
/notebooklm-papers:fetch-papers --json --save papers.jsonProcess multiple papers from a file.
# Process from file
/notebooklm-papers:batch-process ./my-papers.txt
# With audio and custom prefix
/notebooklm-papers:batch-process ./papers.txt --audio --prefix "Project X -"Search arXiv for papers.
# Basic search
/notebooklm-papers:arxiv-search "transformer architecture"
# Filter by category
/notebooklm-papers:arxiv-search "LLM" --category cs.CL
# Search and process top results
/notebooklm-papers:arxiv-search "reinforcement learning" --process 3List your NotebookLM notebooks.
# All notebooks
/notebooklm-papers:list-notebooks
# Filter paper notebooks
/notebooklm-papers:list-notebooks --papers
# Recent only
/notebooklm-papers:list-notebooks --recent 10Generate audio overview for existing notebook.
/notebooklm-papers:generate-audio https://notebooklm.google.com/notebook/abc123
/notebooklm-papers:generate-audio https://notebooklm.google.com/notebook/abc123 --deep-diveThe plugin includes a skill that Claude can automatically invoke when you discuss research papers. It activates when you:
- Mention arXiv papers or IDs
- Ask about creating presentations from papers
- Want to understand or summarize research
- Mention NotebookLM operations
| Operation | Time |
|---|---|
| Source processing | ~30 seconds |
| Slide deck generation | 3-5 minutes |
| Audio overview | 2-3 minutes |
| 5 papers (slides only) | ~25-30 minutes |
| 5 papers (slides + audio) | ~40-50 minutes |
- Start Small: Test with 1-2 papers before running large batches
- Use Parallel Mode:
--parallelflag starts next paper while slides generate - Check Browser: Ensure Chrome is visible and not blocked by popups
- NotebookLM Session: Make sure you're logged into NotebookLM before running
- Patience: Slide generation takes 3-5 minutes per paper
Ensure the Claude in Chrome extension is installed and the browser is open.
- Check you're logged into your Google account
- Refresh NotebookLM and try again
- Clear browser cache if issues persist
- Verify the arXiv ID is valid
- Check if arXiv is accessible
- Some papers may have restricted PDFs
MIT
Contributions welcome! Please submit issues and pull requests to the repository.