Stagehand MCP Server - Local Mode

Stagehand MCP Server - Local Mode

Enables local browser automation using Stagehand without cloud services. Supports navigation, data extraction, and autonomous agent tasks through natural language.

Category
访问服务器

README

Stagehand MCP Server - Local Mode

npm version License: Apache-2.0

Run Stagehand browser automation locally without cloud services. A fork of the official @browserbasehq/mcp-server-browserbase with LOCAL mode support.

Why This Fork?

The official Browserbase MCP server only supports cloud mode, requiring a paid Browserbase subscription. However, Stagehand itself fully supports local browser execution.

This fork unlocks that capability:

Feature Official Version This Fork
Browser execution Browserbase Cloud only Local Headless Chrome
Required credentials BROWSERBASE_API_KEY + PROJECT_ID Only LLM API key
Cost Browserbase subscription Free (bring your own LLM key)
Network requirement Internet required Works offline/intranet
Auto screenshots ✅ After each action

Use Cases

  • Local development - Test browser automation without cloud costs
  • Self-hosted AI agents - Run on your own servers
  • Air-gapped environments - No external cloud dependency
  • CI/CD pipelines - Automated testing without cloud API limits

Quick Start

Using npx (Recommended)

npx stagehand-mcp-local

Add to Claude Code

claude mcp add stagehand-local \
  -e STAGEHAND_ENV=LOCAL \
  -e OPENAI_API_KEY=your_key \
  -- npx stagehand-mcp-local

Add to Cursor / VS Code

Add to your MCP configuration:

{
  "mcpServers": {
    "stagehand-local": {
      "command": "npx",
      "args": ["stagehand-mcp-local"],
      "env": {
        "STAGEHAND_ENV": "LOCAL",
        "OPENAI_API_KEY": "your_openai_key"
      }
    }
  }
}

Installation

npm (Global)

npm install -g stagehand-mcp-local

From Source

git clone https://github.com/weijiafu14/stagehand-mcp-local.git
cd stagehand-mcp-local
pnpm install
pnpm build

Configuration

Environment Variables

Variable Description Default Required
STAGEHAND_ENV Set to LOCAL for local mode BROWSERBASE Yes
OPENAI_API_KEY OpenAI API key - One of these
GEMINI_API_KEY Google Gemini API key - is required
ANTHROPIC_API_KEY Anthropic API key - for Stagehand
HEADLESS Run browser headless true No
SCREENSHOT_ENABLED Enable auto screenshots true No
SCREENSHOT_DIR Screenshot save directory /tmp/stagehand-screenshots No

CLI Options

All original Browserbase MCP server options are supported:

npx stagehand-mcp-local --browserWidth 1920 --browserHeight 1080 --experimental
Flag Description
--browserWidth <width> Browser viewport width (default: 1024)
--browserHeight <height> Browser viewport height (default: 768)
--modelName <model> LLM model for Stagehand (default: gemini-2.0-flash)
--experimental Enable experimental Stagehand features

Available MCP Tools

Once connected, your AI assistant can use these tools:

Tool Description
browserbase_session_create Create a new browser session
browserbase_session_close Close the current session
browserbase_stagehand_navigate Navigate to a URL
browserbase_stagehand_act Perform actions (click, type, etc.)
browserbase_stagehand_extract Extract data from page
browserbase_stagehand_observe Find interactive elements
browserbase_screenshot Take a screenshot
browserbase_stagehand_agent Run autonomous agent task

Auto Screenshots

In LOCAL mode, screenshots are automatically captured after each action for debugging and visualization:

Action performed: Click the login button
[SCREENSHOT:/tmp/stagehand-screenshots/default/1702012345678.jpg]

Parse the screenshot path programmatically:

const match = output.match(/\[SCREENSHOT:(.+?)\]/);
if (match) {
  const screenshotPath = match[1];
  // Use the screenshot...
}

System Requirements

macOS / Windows

Playwright will download Chromium automatically.

Linux (Ubuntu/Debian)

# Install browser dependencies
apt-get update && apt-get install -y \
  chromium libatk1.0-0 libatk-bridge2.0-0 libcups2 \
  libdrm2 libxkbcommon0 libxcomposite1 libxdamage1 \
  libxfixes3 libxrandr2 libgbm1 libasound2 \
  libpango-1.0-0 libcairo2

# Install Playwright browsers
npx playwright install chromium

Docker

FROM node:20

RUN apt-get update && apt-get install -y \
  chromium libatk1.0-0 libatk-bridge2.0-0 libcups2 \
  libdrm2 libxkbcommon0 libxcomposite1 libxdamage1 \
  libxfixes3 libxrandr2 libgbm1 libasound2 \
  libpango-1.0-0 libcairo2 \
  && rm -rf /var/lib/apt/lists/*

ENV PLAYWRIGHT_BROWSERS_PATH=/ms-playwright
RUN npx playwright install chromium

ENV STAGEHAND_ENV=LOCAL

Switching Between Modes

You can switch back to Browserbase cloud mode anytime:

{
  "env": {
    "STAGEHAND_ENV": "BROWSERBASE",
    "BROWSERBASE_API_KEY": "your_key",
    "BROWSERBASE_PROJECT_ID": "your_project_id",
    "GEMINI_API_KEY": "your_gemini_key"
  }
}

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

  1. Fork the repository
  2. Create your feature branch (git checkout -b feature/amazing-feature)
  3. Commit your changes (git commit -m 'Add some amazing feature')
  4. Push to the branch (git push origin feature/amazing-feature)
  5. Open a Pull Request

Credits

This project is a fork of @browserbasehq/mcp-server-browserbase by Browserbase, Inc.

Built with:

License

Apache-2.0 - See LICENSE for details.

Original work: Copyright 2025 Browserbase, Inc. Modifications: Copyright 2025 weijiafu14

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选