Spark
Sign In
MCP Connector

Automate Browser Interactions with Puppeteer

Automate browsers with Puppeteer - navigate, screenshot, click, fill forms, and run JavaScript directly from Claude.

Connect
modelcontextprotocol/servers-archived
Own this? Claim it
82
Spark score
out of 100
Updated 7 months ago
Fresher alternatives ↓
Version 1.0.0
Models
claude
Add to Favorites

Why it matters

Leverage the Puppeteer MCP server to automate browser interactions, enabling AI models to navigate websites, extract data, and perform automated testing.

Outcomes

What it gets done

01

Navigate to specified URLs

02

Interact with web elements (click, fill, hover)

03

Capture screenshots of web pages or elements

04

Execute JavaScript within the browser context

Install

Add it to your toolbox

Run in your project directory:

curl -fsSL https://spark.entire.vc/get/vb-puppeteer | bash

Capabilities

Tools your agent gets

puppeteer_navigate

Navigate to a URL in the browser.

puppeteer_screenshot

Take a screenshot of the current page or specific element.

puppeteer_click

Click on an element identified by CSS selector.

puppeteer_fill

Fill in a form field with text value.

puppeteer_select

Select an option from a dropdown menu.

puppeteer_hover

Hover over an element on the page.

puppeteer_evaluate

Execute JavaScript code in the page context.

Overview

Puppeteer MCP

Puppeteer MCP gives Claude browser automation via Puppeteer - navigation, screenshots, form interaction, and JavaScript execution in the page context - in headless or headed mode. It is an official MCP reference server; its upstream repository is archived. Use it for AI-driven browser automation like scraping, testing, and screenshot documentation. Avoid puppeteer_evaluate with untrusted scripts against pages you don't control.

What it does

Puppeteer MCP enables browser automation through Puppeteer, letting Claude navigate websites, take screenshots, interact with page elements, and extract content, in either headless or headed browser mode. Beyond direct page interaction, it also supports network request interception and PDF generation as broader automation features.

When to use - and when NOT to

Use it when you want Claude to drive a real browser - navigating to a URL, clicking or filling form fields, selecting dropdown options, capturing screenshots (viewport or full page), or executing JavaScript in the page context. It's suited for web scraping, automated testing, screenshot documentation, form automation, and visual regression testing. Do not use puppeteer_evaluate to run untrusted scripts against pages you don't control - it executes arbitrary JavaScript in the page context. This server is one of the official Model Context Protocol reference implementations, now maintained in the modelcontextprotocol/servers-archived repository, which is archived and no longer receiving updates.

Capabilities

  • puppeteer_navigate(url): navigate to a URL.
  • puppeteer_screenshot(options?): capture a screenshot, with name, selector (element-specific), and fullPage options.
  • puppeteer_click(selector): click an element.
  • puppeteer_fill(selector, value): fill a form field.
  • puppeteer_select(selector, value): select a dropdown option.
  • puppeteer_hover(selector): hover over an element.
  • puppeteer_evaluate(script): execute JavaScript in the page context.

Broader features include network request interception and PDF generation.

How to install

npm install -g @modelcontextprotocol/server-puppeteer

Configure Claude Code:

{
  "mcpServers": {
    "puppeteer": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-puppeteer"]
    }
  }
}

For a custom browser path, set the PUPPETEER_EXECUTABLE_PATH environment variable in the server config's env block to point at a specific Chrome binary instead of the bundled one.

Who it's for

Developers who want Claude to automate real browser interactions - scraping pages, testing web flows, capturing screenshots, and filling forms - conversationally.

Source README

The Puppeteer MCP server enables browser automation through Puppeteer, allowing Claude to navigate websites, take screenshots, interact with web elements, and extract content.

Installation

npm install -g @modelcontextprotocol/server-puppeteer

Configuration

Add to your Claude Code settings:

{
  "mcpServers": {
    "puppeteer": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-puppeteer"]
    }
  }
}

With Custom Browser Path

{
  "mcpServers": {
    "puppeteer": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-puppeteer"],
      "env": {
        "PUPPETEER_EXECUTABLE_PATH": "/path/to/chrome"
      }
    }
  }
}

Available Tools

puppeteer_navigate

Navigate to a URL.

puppeteer_navigate(url: string): NavigationResult

puppeteer_screenshot

Take a screenshot of the current page or element.

puppeteer_screenshot(options?: ScreenshotOptions): Screenshot

Options:

  • name - Screenshot identifier
  • selector - CSS selector for element screenshot
  • fullPage - Capture full scrollable page

puppeteer_click

Click on an element.

puppeteer_click(selector: string): ClickResult

puppeteer_fill

Fill in a form field.

puppeteer_fill(selector: string, value: string): FillResult

puppeteer_select

Select an option from a dropdown.

puppeteer_select(selector: string, value: string): SelectResult

puppeteer_hover

Hover over an element.

puppeteer_hover(selector: string): HoverResult

puppeteer_evaluate

Execute JavaScript in the page context.

puppeteer_evaluate(script: string): EvaluationResult

Features

  • Headless and headed browser modes
  • Screenshot capture (viewport or full page)
  • Form interaction (click, fill, select)
  • JavaScript execution in page context
  • Network request interception
  • PDF generation

Usage Example

Claude, navigate to https://example.com, take a screenshot,
then fill in the search form with "MCP servers" and submit it.

Use Cases

  • Web scraping and data extraction
  • Automated testing
  • Screenshot documentation
  • Form automation
  • Visual regression testing

FAQ

Common questions

Discussion

Questions & comments · 0

Sign In Sign in to leave a comment.