Steel Puppeteer
Mirror of
MCP-Mirror
README
Steel Puppeteer
A Model Context Protocol server that provides browser automation capabilities using Puppeteer and Steel. This server enables LLMs to interact with web pages, take screenshots, and execute JavaScript in a real browser environment.
Components
Tools
-
puppeteer_navigate
- Navigate to any URL in the browser
- Inputs:
url
(string, required): URL to navigate totimeout
(number, optional, default: 60000): Navigation timeout in millisecondswaitUntil
(string, optional, default: "domcontentloaded"): When to consider navigation succeeded. Options: "load", "domcontentloaded", "networkidle0", "networkidle2"
-
puppeteer_screenshot
- Capture screenshots of the entire page or specific elements
- Inputs:
name
(string, required): Name for the screenshotselector
(string, optional): CSS selector for element to screenshot
-
puppeteer_click
- Click elements on the page
- Input:
selector
(string, required): CSS selector for element to click
-
puppeteer_fill
- Fill out input fields
- Inputs:
selector
(string, required): CSS selector for input fieldvalue
(string, required): Value to fill
-
puppeteer_select
- Select an element with SELECT tag
- Inputs:
selector
(string, required): CSS selector for element to selectvalue
(string, required): Value to select
-
puppeteer_hover
- Hover elements on the page
- Input:
selector
(string, required): CSS selector for element to hover
-
puppeteer_evaluate
- Execute JavaScript in the browser console
- Input:
script
(string, required): JavaScript code to execute
-
puppeteer_get_content
- Extract content from the current page
- Input:
selector
(string, optional): CSS selector to get content from specific elements. If not provided, returns whole page content
-
puppeteer_scroll
- Scroll the page to trigger lazy-loading
- Inputs:
scrollDelay
(number, optional, default: 100): Delay between scrolls in millisecondsmaxScrolls
(number, optional, default: 50): Maximum number of scrolls
Resources
The server provides access to two types of resources:
-
Console Logs (
console://logs
)- Browser console output in text format
- Includes all console messages from the browser
-
Screenshots (
screenshot://<name>
)- PNG images of captured screenshots
- Accessible via the screenshot name specified during capture
Key Features
- Browser automation with Puppeteer
- Steel integration for browser session management
- Console log monitoring and capture
- Screenshot capabilities
- JavaScript execution
- Basic web interaction (navigation, clicking, form filling)
- Content extraction with token limit handling
- Lazy-loading support through scrolling
- Local and remote Steel instance support
Configuration
Claude Desktop Configuration
To use the Steel Puppeteer server with Claude Desktop, add the following configuration to your Claude Desktop config file (typically located at ~/Library/Application Support/Claude/claude_desktop_config.json
on macOS):
{
"mcpServers": {
"steel-puppeteer": {
"command": "node",
"args": ["path/to/steel-puppeteer/dist/index.js"],
"env": {
"STEEL_LOCAL": "true"
}
}
}
}
Replace "path/to/steel-puppeteer/dist/index.js"
with the actual path to the compiled JavaScript file on your system.
Environment Variables
The Steel Puppeteer server can be configured using the following environment variables:
STEEL_LOCAL
(optional, default: "false"): Set to "true" to use a local Steel instance instead of the cloud service.STEEL_API_KEY
(required only ifSTEEL_LOCAL
is "false"): Your Steel API key for authentication when using the cloud service.STEEL_URL
(optional): The URL of your Steel instance if using a custom deployment.
Server Configuration
If you're running the Steel Puppeteer server directly (not through Claude Desktop), you can set these environment variables in your shell or create a .env
file in the project root. Here's an example .env
file for local usage:
STEEL_LOCAL=true
If you're using the Steel cloud service, your .env
file would look like this:
STEEL_API_KEY=your-steel-api-key
STEEL_LOCAL=false
Running the Server
To start the Steel Puppeteer server:
-
Install dependencies:
npm install
-
Build the project:
npm run build
-
Start the server:
npm start
-
Open Claude Desktop and browse away! The server will start and listen on the specified port (default: 3000).
Troubleshooting
- If you encounter issues with Puppeteer, ensure that you have the necessary dependencies installed on your system. Refer to the Puppeteer troubleshooting guide for more information.
- If using the Steel cloud service, make sure your Steel API key is valid and has the necessary permissions.
- If using a local Steel instance, ensure it's running and accessible at the specified URL (if custom) or at the default local address.
For more detailed configuration options and advanced usage, refer to the Steel documentation and the Puppeteer API reference.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Playwright MCP Server
Provides a server utilizing Model Context Protocol to enable human-like browser automation with Playwright, allowing control over browser actions such as navigation, element interaction, and scrolling.
@kazuph/mcp-fetch
Model Context Protocol server for fetching web content and processing images. This allows Claude Desktop (or any MCP client) to fetch web content and handle images appropriately.
DuckDuckGo MCP Server
A Model Context Protocol (MCP) server that provides web search capabilities through DuckDuckGo, with additional features for content fetching and parsing.
YouTube Transcript MCP Server
This server retrieves transcripts for given YouTube video URLs, enabling integration with Goose CLI or Goose Desktop for transcript extraction and processing.
serper-search-scrape-mcp-server
This Serper MCP Server supports search and webpage scraping, and all the most recent parameters introduced by the Serper API, like location.
The Verge News MCP Server
Provides tools to fetch and search news from The Verge's RSS feed, allowing users to get today's news, retrieve random articles from the past week, and search for specific keywords in recent Verge content.
Tavily MCP Server
Provides AI-powered web search capabilities using Tavily's search API, enabling LLMs to perform sophisticated web searches, get direct answers to questions, and search recent news articles.
mcp-pinterest
A Pinterest Model Context Protocol (MCP) server for image search and information retrieval

Crawlab MCP Server