MCPSERV.CLUB
gpaul-faldin

Puppeteer-Extra MCP Server

MCP Server

Stealth browser automation for LLM web interactions

Stale(50)
2stars
2views
Updated May 23, 2025

About

This MCP server uses Puppeteer-Extra with a stealth plugin to provide human‑like browser automation. It enables LLMs to navigate, interact, capture screenshots, and execute JavaScript while avoiding bot detection.

Capabilities

Resources
Access data sources
Tools
Execute functions
Prompts
Pre-built templates
Sampling
AI model interactions

Puppeteer‑Extra MCP Server Overview

The Puppeteer‑Extra MCP Server equips large language models with a sophisticated, stealthy browser automation layer. By wrapping the popular Puppeteer library together with the Stealth plugin, it allows an AI assistant to navigate, interact, and capture web content in a manner that mimics human browsing behavior. This reduces the likelihood of triggering anti‑automation mechanisms on modern websites, making it ideal for tasks that require reliable web scraping or automated form submission without detection.

At its core, the server exposes a rich set of tools that mirror everyday browser actions. Developers can instruct an LLM to navigate to any URL, click, hover, or fill input fields using simple CSS selectors. The ability to select options from elements and execute arbitrary JavaScript via gives the assistant full control over dynamic pages. Screenshot capabilities are split into two modes: a global page snapshot or targeted element captures, each customizable in size and named for easy retrieval. Additionally, the server streams console logs as a dedicated resource (), enabling real‑time debugging and monitoring of script execution within the browser context.

Real‑world scenarios that benefit from this server are plentiful. For instance, an AI can autonomously log into a web portal, navigate to specific dashboards, and capture screenshots of key metrics for reporting. It can also scrape product listings from e‑commerce sites, filling out search queries and extracting structured data without triggering bot detection. In quality assurance workflows, the assistant can simulate user interactions across multiple pages and report console errors or performance bottlenecks. The stealth mode is especially valuable when dealing with sites that employ sophisticated bot‑detection systems, ensuring that automated runs remain invisible to security measures.

Integration with existing MCP workflows is straightforward. The server registers its tools and resources through the standard MCP manifest, allowing any compliant AI client—such as Claude Desktop—to discover and invoke them via natural language prompts. Developers can chain multiple tool calls within a single conversation, creating complex browsing sequences that the assistant can execute autonomously. Because the server runs headless in production but offers a non‑headless mode for debugging, teams can visually verify interactions during development and then deploy the same logic in a fully automated pipeline.

Unique advantages of this MCP server include its stealth‑enabled automation and the comprehensive set of interaction primitives that mirror real user behavior. Unlike basic headless browsers, Puppeteer‑Extra’s stealth plugin actively masks automation fingerprints, significantly lowering detection rates. The ability to capture both full-page and element screenshots, coupled with live console logging, provides a complete observability stack for web automation tasks. Together, these features make the Puppeteer‑Extra MCP Server a powerful tool for developers looking to blend AI-driven reasoning with reliable, undetectable browser interactions.