Steel Puppeteer

A browser automation service based on Puppeteer and Steel, providing web page interaction capabilities for LLMs

Browser automation Developer tools #Browser automation #Web page interaction #AI agent Local .TypeScript

rating : 2 points

downloads : 9.8K

update time : 2025-04-28

Open Site

What is Steel Puppeteer?

Steel Puppeteer is a Model Context Protocol (MCP) server that provides powerful browser automation capabilities by integrating Puppeteer and Steel. You can use it to navigate web pages, take screenshots, execute JavaScript code, fill out forms, and interact with web page elements.

How to use Steel Puppeteer?

First, install the dependencies and start the server. Then, configure the MCP server address in Claude Desktop to start using it. It supports various functions such as web page navigation, screenshot generation, and JavaScript execution.

Applicable scenarios

Suitable for enterprises and individuals who need web page automation tasks, such as crawler development, web page testing, and data analysis.

Main features

Web page navigation

Allows you to quickly access the target web page by specifying the URL.

Screenshot function

Can capture screenshots of the entire page or a specific area.

JavaScript execution

Runs custom scripts directly in the browser environment.

Lazy loading support

Triggers the lazy loading mechanism by scrolling, facilitating the capture of dynamic content.

Advantages

Powerful browser automation capabilities

Supports multiple operating systems and devices

Easy to integrate into existing systems

Limitations

Requires a certain technical foundation to fully master

May be greatly affected by the network environment

How to use

Install dependencies

Run the npm install command to install the dependencies required for the project.

Build the project

Execute the npm run build command to compile the project files.

Start the server

Use the npm start command to start the Steel Puppeteer service.

Configure Claude Desktop

Add the Steel Puppeteer server address to the configuration file of Claude Desktop.

Usage examples

Example 1: Web page screenshot

Use the puppeteer_screenshot function to get a screenshot of the current page.

Example 2: Form filling

Use the puppeteer_fill function to automatically fill in form fields.

Frequently Asked Questions

How to ensure that Steel Puppeteer works properly?

Can Steel Puppeteer be used remotely?

Related resources

Official documentation

Detailed API reference and user guide.

GitHub repository

Source code and latest version download.

Sample code

Provides some code examples for actual operations.

🚀 Steel Golem Server

A server based on the Model Context Protocol (MCP), leveraging Puppeteer and Steel to provide browser automation capabilities. This server enables Large Language Models (LLMs) to interact with web pages, take screenshots, and execute JavaScript in a real browser environment.

🚀 Quick Start

Install dependencies:
```
npm install
```
Build the project:
```
npm run build
```
Start the server:
```
npm start
```
Open Claude Desktop and start browsing!

✨ Features

Supports advanced browser automation features of Puppeteer.
Integrates the Steel protocol for seamless interaction with LLMs.
Provides flexible configuration options and extensibility.
Supports both local and cloud deployment.

📦 Installation

Environment Variables

The following environment variables can be used to customize the server behavior:

PORT: Specifies the port on which the server runs (default: 3000).
STEEL_API_KEY: The API key for the Steel service.
PUPPETEER_WS_ENDPOINT: The Puppeteer WebSocket endpoint.

Server Configuration

Install dependencies:
```
npm install
```
Build the project:
```
npm run build
```
Start the server:
```
npm start
```

💻 Usage Examples

Components

Tools

puppeteer_navigate
- Navigate to any URL in the browser.
- Input:
  - url (string, required): The URL to navigate to.
  - timeout (number, optional, default: 60000): The navigation timeout in milliseconds.
  - waitUntil (string, optional, default: "domcontentloaded"): The condition to consider the navigation successful. Options: "load", "domcontentloaded", "networkidle0", "networkidle2".
puppeteer_screenshot
- Take a screenshot of a web page or a specific element.
- Input:
  - name (string, required): The name of the screenshot.
  - selector (string, optional): The CSS selector of the element to screenshot.
puppeteer_click
- Click on an element on the web page.
- Input:
  - selector (string, required): The CSS selector of the element to click.
puppeteer_fill
- Fill in a form field.
- Input:
  - selector (string, required): The CSS selector of the form field.
  - value (string, required): The value to fill in.
puppeteer_select
- Select an element in a SELECT tag.
- Input:
  - selector (string, required): The CSS selector of the element to select.
  - value (string, required): The value to select.

Resources

Resources The server provides the following resources to support browser automation operations:
- Screenshot functionality.
- Element interaction (clicking, filling, etc.).
- Browser event listening and control.

🔧 Technical Details

Troubleshooting

If you encounter Puppeteer-related issues, ensure that the necessary dependencies are installed on your system. Refer to the Puppeteer Troubleshooting Guide for more information.
If you are using the Steel cloud service, ensure that your API key is valid and has the necessary permissions.
If you are using a local Steel instance, ensure that it is running and accessible at the specified URL (if customized) or the default local address.

For more detailed configuration options and advanced usage, refer to the Steel documentation and the Puppeteer API reference.