NotebookLM Citation Mapper

A Chrome extension that automatically maps citation numbers to source filenames in Gemini Notebook (formerly Google NotebookLM) and allows you to copy chat text with preserved citations.

Note: Google renamed NotebookLM to Gemini Notebook in July 2026 and moved the app to notebook.google.com. Version 1.3.0 and later support the new address. Older versions stop working because the browser no longer matches them against the page.

Features

• Automatic Citation Mapping: Scans NotebookLM pages and creates a mapping between citation numbers and source documents
• Copy Chat with Citations: Extract the full chat text with citation numbers preserved as [1], [2], etc., with a complete source legend appended
• Citation Mapping Export: Copy just the citation mappings to clipboard
• Smart Page Scanning: Automatically detects new citations as they appear
• Manual Rescan: Refresh the citation mappings on demand

Installation

1. Clone this repository:

   git clone https://github.com/nicremo/notebookLM-citation.git
   cd notebookLM-citation

2. Open Chrome and navigate to chrome://extensions/

3. Enable "Developer mode" (toggle in top right corner)

4. Click "Load unpacked" and select the extension folder from this repository

Usage

1. Navigate to notebook.google.com (formerly notebooklm.google.com)

2. The extension will automatically:

   • Scan for citations in the current notebook
   • Update citations as new content loads

3. Click the extension icon in Chrome to see:

   • Citation Mappings: List of all detected citations
   • 📄 Text mit Quellen kopieren: Copies the entire chat text with citations preserved
   • 📋 Citation Mappings kopieren: Copies just the citation legend
   • 🔄 Rescan Page: Manually refresh the citation mappings

Copy Format

When you click "Text mit Quellen kopieren", the extension copies:

[Your chat text with citations preserved as [1], [2], etc.]

─────────────────────
Quellen:
[1] → Document1.pdf
[2] → Research_Paper.docx
[3] → Notes.txt
...

How It Works

The extension uses three main components:

• content.js: Scans the page for citation markers (using aria-label attributes) and extracts chat text
• popup.js: Provides the popup interface when clicking the extension icon
• background.js: Handles context menu integration and background tasks

Technical Details

• Built for Manifest V3 (latest Chrome Extension API)
• Uses MutationObserver to detect dynamic content changes
• Automatically expands hidden citation lists (... buttons)
• Smart fallback mechanisms for text extraction

Known Limitations

• The citation mapping relies on Gemini Notebook's DOM structure and aria-label attributes
• May break if Google significantly changes the interface. When that happens, run test/selector-contract.js in the DevTools console to see which assumption broke
• Chat text extraction uses heuristics and may occasionally miss content

Changelog

1.3.0

• Works again after Google renamed NotebookLM to Gemini Notebook and moved the app to notebook.google.com
• Fixed collapsed citations being silently skipped, which dropped roughly a third of all citations
• Fixed more_horiz icon text leaking into copied output
• Fixed every paragraph appearing twice in copied output, reported by @ccchan234 in #5
• Fixed the product announcement banner being copied along with the answer
• Fixed a notebook with no chat yet copying hidden accessibility text instead of reporting that there is nothing to copy
• Fixed citation numbers pointing at the wrong source in conversations with more than one answer, contributed by @Jasaj4 in #8. Gemini Notebook numbers citations per response, so the same [1] meant different documents in different answers
• Added a selector contract test so the next interface change is diagnosable in seconds

Thanks to @ejgmd-gh for reporting the outage in #9 and correctly guessing the cause, and to @fothot2 for independently diagnosing it and opening #10.

After updating, reload any Gemini Notebook tab you already had open. Chrome only injects content scripts when a page loads, so tabs opened before the update stay unpatched until you refresh them.

Contributing

Contributions are welcome! If you encounter issues or have ideas for improvements:

1. Open an issue describing the problem or feature request
2. Submit a pull request with your changes

Credits

• Original concept and implementation: @nicremo
• Major refactoring and improvements: @DerSchiman (Ron Schimanski) - Huge thanks for making this project possible by completely rewriting the citation extraction logic and making it actually work!

License

This project is open source and available for anyone to use and modify.