You can connect Claude Desktop directly to NotebookLM using the Model Context Protocol (MCP), which allows Claude to "see" and interact with your notebooks. This transforms Claude into a centralized research command center where it can create new notebooks, add sources, and trigger Audio Overviews through background automation — without you ever leaving the Claude interface.
Most people use NotebookLM and Claude Desktop as separate tools — one for grounded research, the other for open-ended reasoning. The MCP integration collapses that divide. Once connected, Claude can reach into your NotebookLM instance to read sources, create notebooks, compare information across projects, and even generate Audio Overviews on command. You stop tab-switching between tools and start issuing natural-language commands from a single interface.
The real power isn't just convenience. It's that Claude can now reason across your entire NotebookLM workspace. Ask it to find contradictions between two notebooks. Have it pull findings from one project into another. Trigger batch operations — create three notebooks, populate them, and queue audio — in a single prompt. This is what a research command center actually looks like: one place where you think, and the tools respond.
This guide covers the one-time setup, then gives you 30 prompts organized across four capability layers: notebook and source management, cross-notebook research, automation and Audio Overviews, and advanced pipeline workflows. Each prompt is designed to be pasted directly into Claude Desktop once MCP is configured.
The MCP connection is a one-time setup. Once configured, Claude Desktop maintains persistent access to your NotebookLM workspace. Here's how to establish the link and verify it's working.
| Prerequisite | What you need | Where to get it |
|---|---|---|
| Claude Desktop | The desktop application (not the web interface) — MCP connections require the native app for local server communication. | Download from claude.ai/download. Available for macOS and Windows. |
| NotebookLM access | An active Google account with NotebookLM enabled. Works with both free and Plus tiers — Plus gives higher source limits and priority audio generation. | notebooklm.google.com — sign in with your Google account. |
| MCP server package | The NotebookLM MCP server that acts as the bridge between Claude and your notebooks. This is a lightweight local process that Claude communicates with. | Available via the MCP server registry. Check Anthropic's MCP documentation for the latest NotebookLM connector. |
| Node.js runtime | Required to run the MCP server process locally. Version 18 or higher recommended. | nodejs.org — download the LTS version for your operating system. |
Open your terminal and install the NotebookLM MCP server package. This creates a local server process that Claude Desktop will communicate with to access your NotebookLM data. Follow the installation instructions in the package documentation — typically a single npm install or npx command.
Open the Claude Desktop configuration file and add the NotebookLM MCP server entry. This tells Claude where to find the server and how to authenticate with your Google account. The configuration follows the standard MCP server format with your specific credentials and server path.
After saving the config file, restart Claude Desktop completely — the MCP handshake only initializes on startup.
On first launch after configuration, the MCP server will prompt you to authenticate with your Google account. This grants the local server read and write access to your NotebookLM workspace. Complete the OAuth flow in your browser — you'll only need to do this once unless you revoke permissions.
Open a new conversation in Claude Desktop and type: "What notebooks do I currently have in NotebookLM?" If Claude returns a list of your actual notebooks, the MCP connection is live. If it says it doesn't have access, check that the config file is saved correctly and restart Claude Desktop.
Once verified, you can use natural language in Claude Desktop to manage your entire NotebookLM workspace. Create notebooks, add sources, run cross-notebook queries, and trigger Audio Overviews — all without leaving the Claude interface. The prompts below are organized by capability layer to help you get started immediately.
Core commands for creating notebooks, adding sources, and managing your workspace. Paste these directly into Claude Desktop.
Every prompt in this guide plus all prompts across the full category — advanced workflows, specialized use cases, and production-grade templates.
Category Bundle — one-time access
Unlock Category Prompts — $19.99ONE-TIME · 30-DAY GUARANTEE · INSTANT ACCESS
The single highest-leverage habit is to maintain a dedicated "command center" conversation in Claude Desktop that you return to throughout the day. Because Claude remembers context within a session, you can build up complex multi-step workflows: create a notebook in the morning, add sources as you find them during the day, then run cross-notebook analysis and trigger Audio Overviews in the evening — all in one continuous thread.
This beats creating a new Claude conversation for each task, because Claude retains awareness of every notebook you've touched and every operation you've run. It becomes a running research log that also executes commands.
The Model Context Protocol (MCP) is an open standard that lets AI assistants connect to external tools through local server processes. When you configure the NotebookLM MCP server, you're running a lightweight process on your machine that translates Claude's natural-language requests into NotebookLM API calls. Claude doesn't access NotebookLM directly — it talks to the local MCP server, which handles authentication and data retrieval. This means your data never passes through additional third-party infrastructure, and the connection works as long as the local server is running.