# roomi-fields/notebooklm-mcp [Health: Active]

**Category:** 🧠 Knowledge & Memory  
**Repository:** https://github.com/roomi-fields/notebooklm-mcp  
**GitHub Stars:** 218  
**npm Downloads (last month):** 3995  
**Views:** 9  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/roomi-fields-notebooklm-mcp

## Description
Full automation of Google NotebookLM — Q&A with citations, audio podcasts, video, content generation, source management, and notebook library. MCP + HTTP REST API.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "notebooklm-mcp": {
    "command": "npx",
    "args": ["-y","@roomi-fields/notebooklm-mcp"],
    "env": {
      "NOTEBOOKLM_API_KEY": ""
    }
  }
}
```

**Requires environment variables:** `NOTEBOOKLM_API_KEY` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What roomi-fields/notebooklm-mcp does

roomi-fields/notebooklm-mcp exposes Google NotebookLM through two interfaces: an MCP server for agent clients and a 54-endpoint HTTP REST API for tools such as n8n, Zapier, Make, and curl. NotebookLM is also referred to as Gemini Notebook; the project continues to use its original package and repository name.

The service can create and manage notebooks, add sources, ask questions, maintain conversational sessions, and return answers with citations. Citation output can include source names and quoted excerpts, with formats covering no citations, inline citations, footnotes, JSON, and expanded results. It also provides programmatic access to Studio generation, including audio overviews, videos, infographics, reports, presentations, data tables, flashcards, quizzes, and mind maps.

## How it works

The implementation first uses NotebookLM’s internal `batchexecute` RPC interface. A Playwright browser path serves as an automatic fallback when the RPC route cannot be used. These are the same undocumented service interfaces used by the NotebookLM web application rather than an official Google API.

MCP clients can delegate research and content tasks to the server, while ordinary HTTP clients can call the REST API without an agent runtime or MCP client. Multi-account rotation and automatic reauthentication support longer-running workloads. The `vault_batch` workflow writes answers and citations to disk as structured JSON, allowing results to be inspected, searched, compared, or indexed later. Batch processing can resume after an interruption.

## Setup and configuration

Node.js 22 or newer is required. The project can run as an MCP service for clients such as Claude Code, Cursor, and Codex, or as an HTTP service for workflow automation and application backends.

HTTP deployments that need remote access or Docker use require `NOTEBOOKLM_API_KEY`. The HTTP API binds to loopback by default, rejects origins that are not listed, and expects a bearer token when protected access is enabled. A dedicated Google account is recommended for automation, particularly for scheduled or high-volume jobs.

## Tools and capabilities

- Ask NotebookLM questions grounded in notebook sources.
- Return citations in several output formats, including quoted source excerpts.
- Create notebooks, add sources, and manage notebook content.
- Maintain multi-turn sessions with reauthentication when sessions expire.
- Generate and download audio overviews, videos, infographics, reports, presentations, and data tables.
- Generate flashcards, quizzes, and mind maps from notebook material.
- Run large question batches across notebooks and accounts, then persist structured answers to disk.
- Expose the same underlying functionality through MCP and HTTP REST.

## Limitations and notes

This is an unofficial integration and is not affiliated with Google. Its RPC and browser interfaces are undocumented and may change when NotebookLM changes; the project maintains both paths to handle such changes.

NotebookLM’s own quotas still apply, so high-volume work can be constrained by account limits. Account rotation helps distribute batch workloads but does not remove those service quotas. The HTTP API’s security controls are especially relevant when running outside a local development environment: remote access and Docker deployments require an API key, and the service should not be treated as an unauthenticated network endpoint.

_Full upstream README: https://allmcps.com/mcp/roomi-fields-notebooklm-mcp/readme_