NotebookLM MCP Setup Guide

Section 1

Why This Matters

You're building something with an unfamiliar library. You paste the docs into Claude. Claude reads them, burns through your tokens, then confidently writes code that calls functions that don't exist.

You spend an hour debugging. The code was fundamentally broken from line one.

NotebookLM MCP fixes this. Instead of re-reading your files every session, Claude connects directly to Google NotebookLM, where Gemini 2.5 has already indexed your docs. Claude asks it questions. Gets citation-backed answers. Writes correct code.

Zero hallucinations. Minimal tokens. One-time setup.

Approach	Token Cost	Setup	Hallucinations	Answer Quality
Feed docs to Claude	Very high	None	Frequent	Inconsistent
Web search	Medium	None	Occasional	Mixed
Local RAG	Low	Hours	Occasional	Good
NotebookLM MCP	Minimal	5 minutes	Zero	Excellent

Section 2

Setup in 5 Minutes

For Claude Code

Easiest — one command

The One Command:

Terminal

$ claude mcp add notebooklm npx notebooklm-mcp@latest

How to Verify:

1. Type /mcp in Claude Code
2. You should see "notebooklm" in the list of connected servers
3. Ready to go!

For Cursor

Edit ~/.cursor/mcp.json and paste this config:

mcp.json

{
  "mcpServers": {
    "notebooklm": {
      "command": "npx",
      "args": ["notebooklm-mcp@latest"]
    }
  }
}

If you already have other MCP servers configured, add the notebooklm section inside your existing mcpServers object.

Authentication

After installing, tell Claude: "Log me in to NotebookLM"

1. A Chrome window opens automatically
2. Sign in with your Google account
3. Done. Claude is now connected

Other Tools (VS Code, Amp, Codex)

Config instructions for other clients are in the GitHub repo. The pattern is identical to Cursor: same JSON structure, different config file location.

github.com/PleasePrompto/notebooklm-mcp

Section 3

Create Your Knowledge Base

This is where the magic happens. Any docs you add here become zero-hallucination territory for Claude.

1. Create a New Notebook

Go to notebooklm.google.com → click "New notebook"

2. Add Your Docs

NotebookLM accepts almost anything:

PDFs and Google Docs
Markdown files and text files
Websites and documentation URLs
GitHub repositories
YouTube videos

3. Share the Link

Click the ⚙ settings icon → Share → "Anyone with link" → Copy the link

4. Tell Claude

Prompt

"I'm building with [library name]. Here are the docs: [paste your NotebookLM link]"

Section 4

5 Starter Prompts

Copy these prompts and paste them into Claude after setup. Start with #1 or #3. They show the biggest difference.

New Framework Onboarding

I'm starting a new project with [framework/library name]. Here's the documentation in my NotebookLM: [link] Before writing any code, research the docs thoroughly. Ask me any clarifying questions you need. Then outline your implementation plan before touching a single file.

Use Case: Learning new tools, starting greenfield projects, avoiding early wrong assumptions

API Integration Without Hallucinations

I need to integrate [API name]. The full API documentation is in this NotebookLM: [link] Research the authentication method, the correct endpoints, and the exact request/response format before writing anything. Show me the relevant docs sections you found, then build the integration.

Use Case: API integrations, third-party services, anything where hallucinating an endpoint breaks everything

n8n / Workflow Automation

I'm building an n8n workflow that [describe what it should do]. I've added the n8n node documentation to NotebookLM: [link] Research which nodes actually exist for this use case. Don't invent nodes or assume functionality. Then build the workflow JSON with only verified nodes.

Use Case: n8n workflows, Make.com automations, any no-code/low-code tool with specific node names

Internal Codebase Q&A

My codebase documentation is in NotebookLM: [link] Before touching any files, research how [feature/system] works in our codebase. Identify the key files, the data flow, and any patterns we follow. Tell me what you found, then we'll decide what to build.

Use Case: Onboarding to existing projects, understanding legacy code, maintaining architectural consistency

Research Synthesis

The "Wow" Prompt

I've added 10+ sources on [topic] to NotebookLM: [link] Research all of them. Then write a comprehensive brief with: - The main points of agreement across sources - Key debates or contradictions - Specific citations for every major claim - A recommended approach based on the evidence Use NotebookLM's citations throughout so I can verify anything.

Use Case: Deep research, due diligence, synthesizing multiple documentation sources into one coherent brief

Section 5

Pro Tips

1. Use a Dedicated Google Account (Optional)

Not required, but keeps your AI knowledge bases separate from personal notebooks. Easier to manage long-term.

2. One Notebook Per Library or Project

Don't dump everything into one giant notebook. Keep them scoped: one for n8n, one for your Astro docs, one for your internal codebase. Tight scope = better answers.

3. Don't Interrupt the Research Loop

When Claude starts asking follow-up questions to NotebookLM, let it finish. That autonomous research phase is where it builds the context to write correct code. Rushing it defeats the purpose.

4. Use save_notebook to Tag and Auto-Select

Tell Claude to save a notebook link with tags (e.g., "save this as my n8n docs"). Next session, Claude can find and use the right notebook automatically without you pasting the link again.

5. Organize by Topic, Not by Date

Name notebooks by what they contain: "n8n Node Docs", "Stripe API Reference", "Project X Architecture". Date-based names make it hard to know what's in them at a glance.

Zero to Automated

Get the full AI operating system

Guides are a great start. But inside Zero to Automated, you get a done-for-you AI system deployed within 48 hours, plus everything you need to keep building:

Claude Code 101 — 8 lessons, zero to automating real work by Sunday
Ready-to-install Claude Skills — a growing library with 2+ new skills every week
Done-for-you workflows — copy, paste, run. New ones every week.
Weekly live builds — watch real projects get built, ask questions, follow along
Small group coaching — direct access to both founders

See what's inside

$50/mo. Price locked for life at whatever you join at.

NotebookLM MCPSetup Guide