A powerful MCP (Model Context Protocol) server that brings Google NotebookLM into Claude Desktop and Claude Code.
- Research: List and create notebooks
- Content: Add URLs, text, and files as sources
- Generation: Create Podcasts, Videos, Slides, Mind Maps, Infographics, Quizzes, Flashcards, and Reports
- Natural Interaction: Chat directly with your sources using Claude's reasoning
macOS / Linux
# Using curl
curl -LsSf https://astral.sh/uv/install.sh | sh
# Or with Homebrew (macOS)
brew install uvDefault install location: ~/.local/bin/uv
Windows
# Using PowerShell
powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"
# Or with Scoop
scoop install uv
# Or with winget
winget install --id=astral-sh.uv -eDefault install location: %USERPROFILE%\.local\bin\uv.exe
Verify installation:
uv --version# Clone the repository
git clone https://github.com/alfredang/notebooklm-mcp.git
# Navigate to the project folder
cd notebooklm-mcp
# Install dependencies (includes notebooklm-py and fastmcp)
uv syncThis will:
- Create a
.venvvirtual environment - Install notebooklm-py (Python client for NotebookLM API)
- Install fastmcp (MCP server framework)
Note: These dependencies are required for both Claude Desktop and Claude Code.
NotebookLM uses browser-based authentication. You must login once to save your session cookies.
cd notebooklm-mcp
uv run notebooklm loginWhat happens:
- A browser window will open automatically
- Log in to your Google account
- Navigate to NotebookLM if not redirected automatically
- Wait until the terminal displays "Success"
- Close the browser
Verify authentication:
uv run python -c "
from notebooklm import NotebookLMClient
import asyncio
async def test():
client = await NotebookLMClient.from_storage()
async with client:
notebooks = await client.notebooks.list()
print(f'Authenticated! Found {len(notebooks)} notebooks.')
asyncio.run(test())
"You should see: Authenticated! Found X notebooks.
Before configuring Claude, verify the server starts correctly:
cd notebooklm-mcp
uv run python server.pyExpected output:
Starting NotebookLM MCP server...
NotebookLM client initialized successfully
Starting MCP server 'NotebookLM' with transport 'stdio'
Press Ctrl+C (or Cmd+C on Mac) to stop the server after confirming it works.
You'll need two paths for the configuration:
Find your uv path:
macOS / Linux
which uvExample output: /Users/yourname/.local/bin/uv
Windows
where uvExample output: C:\Users\yourname\.local\bin\uv.exe
Find your project path:
macOS / Linux
cd notebooklm-mcp && pwdExample output: /Users/yourname/projects/notebooklm-mcp
Windows
cd notebooklm-mcp; (Get-Location).PathExample output: C:\Users\yourname\projects\notebooklm-mcp
From Claude Desktop (Recommended):
- Open Claude Desktop
- Go to Settings (gear icon) → Developer → Edit Config
- This opens
claude_desktop_config.jsonin your default editor
Or manually:
macOS path
~/Library/Application Support/Claude/claude_desktop_config.json
Windows path
%APPDATA%\Claude\claude_desktop_config.json
Important: Replace
<UV_PATH>and<PROJECT_PATH>with your actual paths from Step 3.1
macOS / Linux Configuration
{
"mcpServers": {
"notebooklm": {
"command": "<UV_PATH>",
"args": [
"--directory",
"<PROJECT_PATH>",
"run",
"python",
"server.py"
]
}
}
}Example with real paths:
{
"mcpServers": {
"notebooklm": {
"command": "/Users/yourname/.local/bin/uv",
"args": [
"--directory",
"/Users/yourname/projects/notebooklm-mcp",
"run",
"python",
"server.py"
]
}
}
}Windows Configuration
{
"mcpServers": {
"notebooklm": {
"command": "<UV_PATH>",
"args": [
"--directory",
"<PROJECT_PATH>",
"run",
"python",
"server.py"
]
}
}
}Example with real paths:
{
"mcpServers": {
"notebooklm": {
"command": "C:\\Users\\yourname\\.local\\bin\\uv.exe",
"args": [
"--directory",
"C:\\Users\\yourname\\projects\\notebooklm-mcp",
"run",
"python",
"server.py"
]
}
}
}Note: On Windows, use double backslashes (
\\) in JSON paths.
| Platform | How to Restart |
|---|---|
| macOS | Press Cmd+Q to fully quit, then reopen |
| Windows | Right-click tray icon → Quit, then reopen |
Look for the hammer icon in the chat input area - this indicates MCP tools are available.
In Claude Desktop, type:
List my NotebookLM notebooks
Claude should use the list_notebooks tool and display your notebooks.
Prerequisites: Complete Steps 1-2 first (install dependencies with
uv syncand authenticate withuv run notebooklm login).
Replace <PROJECT_PATH> with your actual project path:
macOS / Linux
claude mcp add notebooklm -- uv --directory <PROJECT_PATH> run python server.pyExample:
claude mcp add notebooklm -- uv --directory /Users/yourname/projects/notebooklm-mcp run python server.pyWindows
claude mcp add notebooklm -- uv --directory <PROJECT_PATH> run python server.pyExample:
claude mcp add notebooklm -- uv --directory C:\Users\yourname\projects\notebooklm-mcp run python server.pyclaude mcp listYou should see notebooklm in the list.
Start a new Claude Code session:
claudeThen ask:
List my NotebookLM notebooks
Once configured, use natural language commands in Claude Desktop or Claude Code:
| Task | Example Command |
|---|---|
| List notebooks | "Show me all my NotebookLM notebooks" |
| Create notebook | "Create a new notebook called 'Research Project'" |
| Add URL source | "Add this URL to my notebook: https://example.com/article" |
| Generate podcast | "Generate a podcast for notebook ID xyz123" |
| Create slides | "Make a slide deck from my 'Research Project' notebook" |
| Generate mind map | "Create a mind map for notebook abc456" |
| Create quiz | "Generate a quiz based on my notebook sources" |
| Make flashcards | "Create study flashcards from this notebook" |
| Tool | Description |
|---|---|
list_notebooks |
List all notebooks in your account |
create_notebook |
Create a new notebook |
add_source_url |
Add a website URL as a source |
add_source_text |
Add raw text as a source |
ask_notebook |
Ask a question based on notebook sources |
get_notebook_summary |
Get summary and key insights |
generate_audio_overview |
Generate a podcast-style audio |
generate_video_overview |
Generate a video overview |
generate_slide_deck |
Generate PowerPoint-style slides |
generate_mind_map |
Generate an interactive mind map |
generate_infographic |
Generate a visual infographic |
generate_quiz |
Generate quiz questions |
generate_flashcards |
Generate study flashcards |
generate_summary_report |
Generate a briefing document |
generate_data_table |
Extract data into a table |
Cause: Claude Desktop can't find uv because it doesn't inherit your shell's PATH.
Solution: Use the full absolute path to uv in the config (see Step 3.1).
macOS / Linux
Add to your shell profile (~/.zshrc or ~/.bashrc):
export PATH="$HOME/.local/bin:$PATH"Then reload:
source ~/.zshrc # or source ~/.bashrcWindows
Add to your PATH:
- Open System Properties → Environment Variables
- Under "User variables", edit
Path - Add:
%USERPROFILE%\.local\bin - Restart your terminal
Cause: Invalid JSON in config file or Claude not restarted properly.
Solution:
- Validate your JSON at https://jsonlint.com/
- Ensure no trailing commas in the JSON
- Fully quit and reopen Claude Desktop
Cause: Server started before authentication was complete.
Solution:
- Run
uv run notebooklm loginfirst - Restart Claude Desktop or re-add the MCP server in Claude Code
macOS
# View recent logs
tail -100 ~/Library/Logs/Claude/mcp*.log
# Or open in Finder
open ~/Library/Logs/Claude/Windows
# View logs folder
explorer "$env:APPDATA\Claude\logs"If issues persist:
claude mcp remove notebooklm
claude mcp add notebooklm -- uv --directory <PROJECT_PATH> run python server.pyTo update the NotebookLM library:
cd notebooklm-mcp
uv sync --upgradenotebooklm-mcp/
├── server.py # MCP server implementation
├── pyproject.toml # Project dependencies
├── README.md # This file
├── SKILL.md # Claude Code skill definition
└── .venv/ # Virtual environment (auto-created)
MIT License