OneClickLM

MCP其他

提供方 CRtheHILLS · 包名 oneclicklm

其他 stdio · 本地进程 API Key 社区 / 第三方 ClaudeCursorVS CodeWindsurf
查看接入方式

OneClickLM is a zero-config, auto-healing NotebookLM MCP server for AI assistant integration, enabling natural-language listing, querying, creating, and managing of NotebookLM notebooks and sources.

一句话选型结论

按资料字段整理

OneClickLM 属于其他类 MCP 服务器,stdio · 本地进程,API Key,可直接接入 Claude、Cursor、VS Code 等,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。

分类其他 传输方式stdio · 本地进程 鉴权API Key 部署方式本地部署
GitHub 星标 3 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

其他
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 MCP server NotebookLM integration TypeScript AI notebook management zero-config setup auto-healing authentication natural language query

支持客户端

适用场景

Local DeploymentMCP serverNotebookLM integration

安装与配置

stdio · 本地进程 API Key
客户端在本机拉起子进程,数据不出本机
npx oneclicklm

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 node,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
接入外部数据

把外部系统的数据接进对话。

自动化重复操作

把固定步骤交给模型跑。

补充工具能力

给模型补上它原本没有的能力。

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、Cursor、VS Code,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API Key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

  • 需要在其他场景里让模型直接调用外部能力的人。
  • 已经在用 Claude、Cursor 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
OneClickLM当前 未标注 stdio · 本地进程 本地部署 有 2026-03-08
gemini-cli 未标注 stdio · 本地进程 本地部署 有 2026-04-13
ext-apps 未标注 stdio · 本地进程 本地部署 有 2026-04-19
northcinder 未标注 stdio · 本地进程 本地部署 有 2026-08-23
fastctx 未标注 stdio · 本地进程 本地部署 有 2026-08-23
quickstart-resources 未标注 stdio · 本地进程 本地部署 有 2026-04-12

服务器档案

其他
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CRtheHILLS上游标注的维护方 CRtheHILLS
分类其他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API Key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oneclicklm · node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2026-03-08上游最后核验时间

详细介绍

7.7 KB
来自项目仓库与上游目录的原文,未做删改

OneClickLM

The NotebookLM MCP server that actually works.

Auto-healing auth · Zero config · One command · 6 powerful tools

Quick Start ·
The Problem ·
Tools ·
IDE Setup ·
Comparison


Hi! I'm CR 👋

I built OneClickLM because I was tired of every NotebookLM MCP server breaking on me every few days.

After connecting NotebookLM to a 6-domain AI chatbot platform (BEYOND HUMAN), I hit literally every failure mode — expired tokens, Chrome crashes, concurrent query timeouts, you name it. So I built the tool I wished existed.

OneClickLM just works. Login once, forget about it forever.


💥 The Problem

Every NotebookLM MCP server on GitHub shares the same fatal flaw: they break within days.

Google silently rotates auth tokens, and no one handles it.

What happens What you see
build_label expires (every 1-2 weeks) 🔴 400 Bad Request, zero explanation
CSRF token goes stale 🔴 Queries silently fail
Chrome already running 🔴 Login crashes with CDP conflict
Two queries at once 🔴 Timeout cascade, server dies
Python + pipx + virtualenv 🔴 30 minutes just to install
Connection drops 🔴 Queries hang forever

Sound familiar? Yeah, I went through all of this. So you don't have to.


✅ The Solution

Feature How it works
Auto-healing auth Detects expired tokens → refreshes automatically → you never notice
Zero config npx oneclicklm login once, then it just works. Forever.
No Chrome conflicts Uses system Chrome for login, then pure HTTP. No CDP, no conflicts.
Smart queue Serializes concurrent calls. No timeouts, no crashes.
6 powerful tools List, query, create notebooks + add sources. Everything you need.
TypeScript native No Python, no virtualenv, no pipx. Just Node.js.

🚀 Quick Start

Step 1: Login (one time only)```bash
npx oneclicklm login


A browser opens → sign in with Google → done. Your cookies are saved locally (~/.oneclicklm/).

**Step 2: Add to your MCP client**```json
{
  "mcpServers": {
    "notebooklm": {
      "command": "npx",
      "args": ["oneclicklm"]
    }
  }
}

Step 3: Start talking to your notebooks!```
"List my NotebookLM notebooks"
"What are the key findings in my Research notebook?"
"Create a new notebook called 'Project Alpha'"
"Add this URL to my notebook: https://example.com/paper.pdf"


That's it. No API keys, no profiles, no metadata.json, no build_label hunting.

---

## 🛠 Available Tools

| Tool | What it does | Example prompt |
|:---|:---|:---|
| `notebook_list` | List all your notebooks | "Show me my NotebookLM notebooks" |
| `notebook_get` | Get notebook details + sources | "What sources are in my Research notebook?" |
| `notebook_query` | Ask questions, get cited answers | "What does my notebook say about X?" |
| `notebook_create` | Create a new notebook | "Create a notebook called 'Q1 Report'" |
| `source_add` | Add URL, YouTube, or text sources | "Add this article to my notebook" |
| `source_list` | List all sources + their status | "Show sources in my Project notebook" |

> **Pro tip:** notebook_query uses the same AI (Gemini) as the NotebookLM web app — your answers are grounded in your actual sources with zero hallucination.

---

## 💻 IDE Setup

🟢 Cursor

Add to `~/.cursor/mcp.json`:```json
{
  "mcpServers": {
    "notebooklm": {
      "command": "npx",
      "args": ["oneclicklm"]
    }
  }
}

🟣 Claude Code (CLI)```bash
claude mcp add notebooklm -- npx oneclicklm


🔵 VS Code (Copilot)

Add to `.vscode/mcp.json`:```json
{
  "servers": {
    "notebooklm": {
      "command": "npx",
      "args": ["oneclicklm"]
    }
  }
}

🟡 Windsurf

Add to ~/.windsurf/mcp.json:```json
{
"mcpServers": {
"notebooklm": {
"command": "npx",
"args": ["oneclicklm"]
}
}
}


---

## 🔍 How OneClickLM Compares

| Feature | OneClickLM | notebooklm-mcp (Python) | notebooklm-mcp-cli |
|:---|:---:|:---:|:---:|
| Auto token refresh | ✅ | ❌ | ❌ |
| Auto build_label update | ✅ | ❌ | ❌ |
| Auto-reconnect on failure | ✅ | ❌ | ❌ |
| Request queue (no crashes) | ✅ | ❌ | ❌ |
| Zero-config install | ✅ `npx` | ❌ pipx + profiles | ❌ pipx + profiles |
| Language | TypeScript | Python | Python |
| Chrome conflicts | None | Common | Common |
| Human-readable errors | ✅ | ❌ | ❌ |
| Create notebooks via MCP | ✅ | ❌ | ✅ |
| Add sources via MCP | ✅ | ❌ | ✅ |

---

## ⚙️ Configuration

OneClickLM works with **zero configuration**. But if you need it:```bash
# Custom config directory
ONECLICKLM_DIR=~/.my-config npx oneclicklm

# Debug logging (see what's happening under the hood)
ONECLICKLM_LOG=debug npx oneclicklm

# Custom request timeout (default: 30s, queries may need longer)
ONECLICKLM_TIMEOUT=60000 npx oneclicklm

# Custom Chrome path
CHROME_PATH=/path/to/chrome npx oneclicklm login

Config is stored in ~/.oneclicklm/:```
~/.oneclicklm/
├── cookies.json # Google auth cookies (auto-managed)
└── tokens.json # CSRF + session + build_label (auto-refreshed)


---

## 🧠 How It Works (for the curious)

OneClickLM speaks the same language as the NotebookLM web app — Google's internal `batchexecute` RPC protocol.

Here's what happens when you ask a question:```
You: "What are the key findings?"
 ↓
MCP Client (Cursor/Claude/VS Code) sends tool call
 ↓
OneClickLM checks tokens → auto-refreshes if stale
 ↓
Fetches source IDs from notebook (RPC: rLM1Ne)
 ↓
Sends query via streaming endpoint (GenerateFreeFormStreamed)
 ↓
Parses streaming response → extracts answer
 ↓
Returns grounded, cited answer to your AI

The magic is in the auto-healing auth:
1. On startup: loads saved cookies + cached tokens
2. If tokens expired (>1 hour): fetches NotebookLM page → extracts fresh SNlM0e (CSRF) + FdrFJe (session) + build_label
3. If cookies expired (~30 days): prompts for re-login
4. If API returns 400/401: auto-refreshes tokens and retries once
5. All requests serialized through a queue → no concurrent crash

> [!NOTE]
> OneClickLM uses Google's internal web protocol, not an official API. This means it works with your existing Google account — no API keys, no billing, no quotas. Google AI Pro subscribers get full NotebookLM access.


🔧 Troubleshooting

Login window doesn't appear

Make sure Chrome/Chromium is installed. OneClickLM detects it automatically on macOS, Windows, and Linux.```bash

If using a non-standard Chrome location:

CHROME_PATH=/path/to/chrome npx oneclicklm login


Getting 400/401 errors

OneClickLM auto-refreshes tokens, but if it persists:```bash
# Force token refresh
npx oneclicklm refresh

# Nuclear option: re-login
npx oneclicklm login

Queries are slow (~15-30 seconds)

This is normal. NotebookLM processes your query against all uploaded sources in real-time (this is the Gemini model working). The NotebookLM web app has similar response times. For faster results, use notebooks with fewer sources.

Check if everything is working```bash
npx oneclicklm status


This verifies cookies, extracts fresh tokens, and tells you if anything is wrong.

---

## 🤝 Contributing

Contributions welcome! Bug reports, feature requests, and PRs are all appreciated.```bash
git clone https://github.com/CRtheHILLS/OneClickLM.git
cd OneClickLM
npm install
npm run build
node dist/src/index.js  # Run locally

⭐ Star History

If OneClickLM saved you from NotebookLM auth hell, consider starring the repo!


License

MIT © CRtheHILLS


Built with frustration, fixed with determination. 🔥

Born from BEYOND HUMAN — a 6-domain AI chatbot platform powered by NotebookLM.

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
OneClickLM 怎么安装?

在客户端按下面的安装命令接入:npx oneclicklm。对应的包名是 oneclicklm。装完在客户端配置里引用同一个包或端点。

OneClickLM 怎么在 Claude、Cursor、VS Code 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。

OneClickLM 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。

OneClickLM 需要 API Key 吗?

资料标注需要API Key认证,使用前准备好对应凭据,并按最小权限配置。

OneClickLM 支持哪些客户端?

资料标注支持 Claude、Cursor、VS Code、Windsurf、Gemini CLI、Zed;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。

OneClickLM 能提供哪些能力?

资料显示能力标签包括 Local Deployment、MCP server、NotebookLM integration、TypeScript、AI notebook management、zero-config setup。具体工具清单以官方文档为准。

OneClickLM 需要什么运行环境?

运行时为 node;属于本地进程,需要在本机装好运行环境。

OneClickLM 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 2026-03-08。