The Model Context Protocol (MCP) is an open standard by Anthropic that enables AI assistants and agents to connect with external tools, APIs, and data sources through a unified interface. The ecosystem has grown to over 10,000 public servers since its November 2024 launch — spanning browser automation, databases, cloud infrastructure, DevOps, productivity, security, and dozens of other domains. A 2026-07-28 release candidate introduces a fully stateless protocol core, the MCP Apps extension for server-rendered UIs, and an official Extensions framework — making MCP the de facto USB-C of AI tooling. The key mental model is that each MCP server acts as a bridge between an LLM and a specific external system, exposing structured tools (functions the model can call), resources (data the model can read), and prompts (templates for common tasks) — so the most important decision is not which AI you use, but which MCP servers you connect it to.
What This Cheat Sheet Covers
This topic spans 15 focused tables and 280 indexed concepts. Below is a complete table-by-table outline of this topic, spanning foundational concepts through advanced details.
Table 1: Browser Automation & Desktop Control
Browser and desktop MCP servers give AI agents eyes and hands — the ability to navigate real pages, fill forms, extract structured data, and control OS-level UI without screenshots. Choosing between cloud-hosted headless solutions (Browserbase, Hyperbrowser) and locally installed drivers (Playwright, Puppeteer, Selenium) depends on whether you need anti-bot bypass or just programmatic control.
| Server | Example | Description |
|---|---|---|
npx /mcp | • Official Microsoft browser automation • 28,900+ ★ • uses accessibility tree snapshots for structured interaction without screenshots | |
npx /mcp-server-browserbase | • Cloud-hosted headless browser with anti-detection • Stagehand integration for AI-native scraping • managed sessions | |
npx -y /server-puppeteer | • Official reference MCP server • lightweight Puppeteer-based browser control • page navigation, screenshots, form filling, JS execution | |
pip install browser-use | • 6,100+ ★ • full browser agent framework • vision + HTML extraction • multi-tab support | |
npx /mcp | • AI-optimized cloud browser • built-in CAPTCHA solving and stealth mode • session management | |
npx agentql-mcp | • Natural language selectors for web elements • AI-powered semantic querying of page content instead of CSS/XPath | |
npx /stagehand | • AI-native web browsing SDK • act(), extract(), observe() primitives• natural language actions |