NotebookLM-MCP

MCP浏览器工具

提供方 kkaminsk

浏览器工具 stdio · 本地进程 会话登录 社区 / 第三方 Claude DesktopClaude
查看接入方式

A Python-based Model Context Protocol (MCP) server that bridges AI agents with Google NotebookLM through browser automation, enabling MCP-compatible clients to utilize NotebookLM's powerful Retrieval-Augmented Generation (RAG) capabilities.

一句话选型结论

按资料字段整理

NotebookLM-MCP 属于浏览器工具类 MCP 服务器,stdio · 本地进程,会话登录,可直接接入 Claude Desktop、Claude,有可直接执行的安装命令。接入前先确认权限范围;涉及写入、删除或付款的操作,建议先在测试环境验证。

分类浏览器工具 传输方式stdio · 本地进程 鉴权会话登录 部署方式本地部署
GitHub 星标 2 仓库热度,无公开仓库时显示未标注
工具数 未标注 暴露给模型的工具数量
资源 / 提示词 0 / 0 资源与提示词模板数量
维护方 社区 / 第三方 核验等级 仅资料核对

工具与能力

浏览器工具
数量来自来源目录的标注,具体工具清单以官方文档为准
工具数量未标注暴露给模型可调用的工具个数
资源数量0可读取的资源条目
提示词数量0内置提示词模板数量

能力标签

Local Deployment Browser Automation RAG Python AI Agents NotebookLM

支持客户端

适用场景

Local DeploymentBrowser AutomationRAG

安装与配置

stdio · 本地进程 会话登录
客户端在本机拉起子进程,数据不出本机
uv --directory C:\path\to\notebooklm-mcp run notebooklm-mcp serve

这条命令只是入口,具体启动参数请按官方文档补齐,不要直接照搬到别的服务器。

  1. 1
    准备运行环境

    stdio 服务在本地起进程,需要先装好 python,再按安装命令拉取依赖。

  2. 2
    安装服务

    复制上面的安装命令执行;命令里带占位符的位置,换成自己的密钥或路径。

  3. 3
    写入客户端配置

    在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,让新配置生效。

  4. 4
    验证是否可用

    连接成功后随便调一个只读工具,确认能返回结果,再接入正式流程。

典型使用场景

3 类
按分类与标签整理的常见用法,具体以官方文档为准
自动操作网页

打开页面、点击、填表,把浏览器动作交给模型执行。

抓取结构化数据

把页面内容整理成表格或 JSON,省掉手工复制粘贴。

关键流程回归

按步骤检查页面是否正常,回归时少漏环节。

选择建议

选之前先看这几条
按条目字段给出的判断,不替你决定用不用

值得选的地方

  • 有现成的安装命令,不用自己找入口,照着跑就能起服务。
  • 已适配 Claude Desktop、Claude,换客户端不用重新研究配置。

先想清楚的地方

  • stdio 方式要求本地装好运行时(如 Node.js),换机器要重新装一遍。
  • 使用前要准备好会话登录对应的凭据,并按最小权限配置。
  • 资料没标注工具数量,实际能力要装好后确认。
  • 版本更新与行为细节以官方仓库为准。

更适合

  • 需要在浏览器工具场景里让模型直接调用外部能力的人。
  • 已经在用 Claude Desktop、Claude 的用户,接入路径最短。
  • 想先快速试一下、能接受在本地或自有环境跑服务的人。

不太适合

  • 完全不想在本地装环境的人。
  • 不方便管理密钥或第三方授权的人。
  • 需要厂商级 SLA 和技术支持的生产环境。

同类对比

6 个同类
同分类里挑最接近的几条,字段以各自详情页为准
MCP 服务器工具数传输方式部署方式安装命令最后核验
NotebookLM-MCP当前 未标注 stdio · 本地进程 本地部署 有 2025-12-31
Playwright MCP Server 21 stdio · 本地进程 本地部署 有 2026-05-18
Fetch 9a2682de 1 SSE · 服务端推送 支持远程部署 有 2026-05-18
Browser 738d52bc 未标注 stdio · 本地进程 本地部署 有 2026-05-17
MCP 40600ece 未标注 stdio · 本地进程 本地部署 有 2026-05-17
Skyvern MCP 未标注 stdio · 本地进程 本地部署 有 2026-05-17

服务器档案

浏览器工具
字段来自项目仓库与公开索引,中文说明由本站标注
提供方kkaminsk上游标注的维护方 kkaminsk
分类浏览器工具本站按上游分类展示
连接方式stdio · 本地进程客户端在本机拉起子进程,数据不出本机
鉴权方式会话登录远程服务建议用最小权限凭据
维护方社区 / 第三方核验等级 仅资料核对
包名 / 运行时未标注 · python本地安装时用到的包标识与运行时
部署类型本地部署上游给出的部署形态
最近核验2025-12-31上游最后核验时间

详细介绍

8.9 KB
来自项目仓库与上游目录的原文,未做删改

NotebookLM MCP Server

Version 0.5 Alpha | Basic functionality implemented, requires further testing

A Python-based Model Context Protocol (MCP) server that bridges AI agents with Google NotebookLM through browser automation. This allows MCP-compatible clients like Claude Desktop to interact with NotebookLM's powerful RAG (Retrieval-Augmented Generation) capabilities.

> Warning: This is alpha software. It has basic functionality but needs further testing. Use at your own risk. Automating Google services may violate their Terms of Service.

Features

  • Notebook Management: Create notebooks, list existing notebooks
  • Source Management: Add sources (URLs, text content), list sources in notebooks
  • Query Interface: Ask questions against notebook sources with citation support
  • Audio Overview: Generate podcast-style audio discussions from notebook content
  • Session Persistence: Human-in-the-loop authentication with persistent Chrome profile
  • Stealth Automation: Uses Patchright for bot detection evasion

Architecture```

┌─────────────────┐ ┌──────────────────┐ ┌─────────────────┐
│ MCP Client │────▶│ NotebookLM MCP │────▶│ Google │
│ (Claude, etc) │ │ Server │ │ NotebookLM │
└─────────────────┘ └──────────────────┘ └─────────────────┘
│
▼
┌──────────────┐
│ Patchright │
│ (Browser) │
└──────────────┘


The server uses:
- **FastMCP**: For MCP protocol implementation
- **Patchright**: Stealth browser automation (Playwright fork with anti-detection)
- **Persistent Chrome Profile**: For session management without repeated logins

## Prerequisites

- Python 3.11 or higher
- Google Chrome browser installed
- A Google account with access to NotebookLM
- uv package manager (recommended) or pip

## Installation

### Using uv (Recommended)```bash
# Clone the repository
git clone https://github.com/your-org/notebooklm-mcp.git
cd notebooklm-mcp

# Install dependencies
uv sync

# Install Patchright browsers
uv run patchright install chromium

Using pip```bash

Clone the repository

git clone https://github.com/your-org/notebooklm-mcp.git
cd notebooklm-mcp

Create virtual environment

python -m venv .venv
source .venv/bin/activate # On Windows: .venv\Scripts\activate

Install package

pip install -e .

Install Patchright browsers

patchright install chromium


## Configuration

The server stores configuration and session data in `~/.notebooklm-mcp/` by default:```
~/.notebooklm-mcp/
├── config.toml          # Optional configuration file
├── chrome_profile/      # Persistent browser profile (contains session)
└── logs/                # Log files (if configured)

Configuration Options

Create ~/.notebooklm-mcp/config.toml to customize settings:```toml
[server]
log_level = "INFO" # DEBUG, INFO, WARNING, ERROR
cache_enabled = true # Enable response caching
browser_headless = true # Run browser in headless mode
browser_timeout_ms = 30000 # Browser operation timeout

[browser]
page_pool_min = 1 # Minimum pages in pool
page_pool_max = 3 # Maximum pages in pool
idle_timeout = 300 # Close browser after idle (seconds)


## Usage

### 1. Authentication

Before using the server, you must authenticate with your Google account:```bash
# Using uv
uv run notebooklm-mcp auth

# Or if installed via pip
notebooklm-mcp auth

This opens a visible Chrome browser where you can:
1. Log in to your Google account
2. Complete any 2FA challenges
3. The browser closes automatically after successful authentication

Your session is saved to the Chrome profile directory for future use.

2. Verify Session

Check if your session is still valid:```bash
uv run notebooklm-mcp verify


### 3. Start the MCP Server```bash
uv run notebooklm-mcp serve

The server runs using stdio transport for MCP communication.

4. Other CLI Commands```bash

Show status and configuration

uv run notebooklm-mcp status

Clear saved session (requires re-authentication)

uv run notebooklm-mcp clear


## Claude Desktop Integration

Add the server to your Claude Desktop configuration:

### Windows

Edit `%APPDATA%\Claude\claude_desktop_config.json`:```json
{
  "mcpServers": {
    "notebooklm": {
      "command": "uv",
      "args": [
        "--directory",
        "C:\\path\\to\\notebooklm-mcp",
        "run",
        "notebooklm-mcp",
        "serve"
      ]
    }
  }
}

macOS/Linux

Edit ~/.config/claude/claude_desktop_config.json:```json
{
"mcpServers": {
"notebooklm": {
"command": "uv",
"args": [
"--directory",
"/path/to/notebooklm-mcp",
"run",
"notebooklm-mcp",
"serve"
]
}
}
}


## Available MCP Tools

Once connected, the following tools are available to MCP clients:

### `list_notebooks`
List all notebooks in your NotebookLM account.

### `create_notebook`
Create a new notebook with optional initial sources.
- **Parameters**: `title` (required), `initial_sources` (optional list of URLs)

### `add_source`
Add a source to a notebook.
- **Parameters**: `notebook_id`, `source_type` (url, text, google_doc, google_slide), `content`, `title` (optional)

### `list_sources`
List all sources in a notebook.
- **Parameters**: `notebook_id`

### `query_notebook`
Ask a question against notebook sources.
- **Parameters**: `notebook_id`, `query`, `use_chat_history` (optional)

### `generate_audio_overview`
Generate a podcast-style audio discussion.
- **Parameters**: `notebook_id`, `custom_instructions` (optional)

### `health_check`
Check server and browser health status.

## Running Workflows

### Example: Create Notebook and Generate Audio```python
import asyncio
from patchright.async_api import async_playwright

async def create_notebook_workflow():
    playwright = await async_playwright().start()

    context = await playwright.chromium.launch_persistent_context(
        user_data_dir="~/.notebooklm-mcp/chrome_profile",
        headless=False,
        channel="chrome",
    )

    page = context.pages[0] if context.pages else await context.new_page()

    # Navigate to NotebookLM
    await page.goto("https://notebooklm.google.com/")

    # ... interact with NotebookLM

    await context.close()
    await playwright.stop()

asyncio.run(create_notebook_workflow())

Using the MCP Tools Programmatically

When connected via MCP, you can invoke tools like:```
User: Create a new notebook called "Research Notes" and add this URL as a source: https://example.com/article

Claude: I'll create the notebook and add the source for you.
[Calls create_notebook with title="Research Notes"]
[Calls add_source with the URL]

The notebook has been created with ID abc123 and the source is being processed.


## Security Considerations

- **Chrome Profile**: Contains your Google session credentials. Keep `~/.notebooklm-mcp/chrome_profile/` secure and never commit it to version control.
- **Local Execution**: The server is designed for local use. If deploying remotely, use secure tunnels (SSH) as MCP doesn't provide encryption.
- **Rate Limiting**: Avoid rapid automated requests to prevent account flagging.

## Known Limitations

- **No Official API**: This uses browser automation as NotebookLM lacks a public API
- **DOM Dependency**: May break if Google updates NotebookLM's interface
- **Session Expiry**: Sessions may expire and require re-authentication
- **PDF Upload**: Not yet implemented (URL and text sources work)
- **Alpha Status**: Basic functionality works but edge cases need testing

## Troubleshooting

### "Authentication expired" error
Run `uv run notebooklm-mcp auth` to re-authenticate.

### "net::ERR_NAME_NOT_RESOLVED" error
Clear the Chrome profile and re-authenticate:```bash
uv run notebooklm-mcp clear
uv run notebooklm-mcp auth

Browser not launching

Ensure Chrome is installed and Patchright browsers are set up:```bash
uv run patchright install chromium


## Development```bash
# Install dev dependencies
uv sync --all-extras

# Run tests
uv run pytest

# Run linter
uv run ruff check src/

# Format code
uv run ruff format src/

License

MIT License - see LICENSE for details.

Disclaimer

This is an unofficial integration. NotebookLM is a trademark of Google. This project is not affiliated with, endorsed by, or sponsored by Google. Use of this software may violate Google's Terms of Service. Users assume all risks associated with using this software.

常见问题

8 问
答案由本站依据条目字段整理,不引用上游原文
NotebookLM-MCP 怎么安装?

在客户端按下面的安装命令接入:uv --directory C:\path\to\notebooklm-mcp run notebooklm-mcp serve。装完在客户端配置里引用同一个包或端点。

NotebookLM-MCP 怎么在 Claude Desktop、Claude 里配置?

它的传输方式是stdio · 本地进程,在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,配置才会生效。

NotebookLM-MCP 收费吗?

本站不代收费用,也没有标注具体价格。是否收费取决于服务本身和背后的接口额度,正式使用前按官方说明确认。

NotebookLM-MCP 需要 API Key 吗?

资料标注需要会话登录认证,使用前准备好对应凭据,并按最小权限配置。

NotebookLM-MCP 支持哪些客户端?

资料标注支持 Claude Desktop、Claude;其他客户端只要能配置 MCP Server,也能用同样的命令或地址接入。

NotebookLM-MCP 能提供哪些能力?

资料显示能力标签包括 Local Deployment、Browser Automation、RAG、Python、AI Agents、NotebookLM。具体工具清单以官方文档为准。

NotebookLM-MCP 需要什么运行环境?

运行时为 python;属于本地进程,需要在本机装好运行环境。

NotebookLM-MCP 是本地还是远程?

它是本地进程(stdio),由客户端在本机拉起。传输方式为stdio · 本地进程。

热门专题

按标签挑服务器
全部专题 →

服务器信息以作者仓库与官方文档为准;分类、传输方式与鉴权方式的中文说明为中文整理,最近核验 2025-12-31。