MCP + Playwright: Complete Guide
AI agents are becoming more powerful, but they need a safe and structured way to interact with the real world—especially web browsers. This is where MCP (Model Context Protocol) combined with Playwright comes in.
In this blog, you’ll learn:
- What MCP is
- What Playwright is
- Why they work perfectly together
- How to build an MCP Playwright server
- Line-by-line code explanation
1. What is MCP (Model Context Protocol)?
MCP is a protocol that allows AI models to call tools, use external services, receive structured responses, and operate securely. Think of MCP as a bridge between AI models and real-world tools like browsers, files, and APIs.
2. What is Playwright?
Playwright is a browser automation library created by Microsoft. It allows you to open websites, click buttons, fill forms, take screenshots, and scrape content across Chromium, Firefox, and WebKit.
3. Why Use MCP with Playwright?
Without MCP
- AI automation is unsafe
- No standard tool interface
- Hard to control permissions
With MCP + Playwright
- Secure browser automation
- Standardized tool interface
- Works with any AI model
4. Step-by-Step Implementation
Step 1: Project Setup
$ mkdir mcp-playwright
$ cd mcp-playwright
$ npm init -yStep 2: Install Dependencies
$ npm install @modelcontextprotocol/sdk playwright
$ npx playwright installStep 3: Basic MCP Server Setup
Create a file called index.js and add the following:
import { Server } from "@modelcontextprotocol/sdk/server/index.js";
import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
import { chromium } from "playwright";
const server = new Server(
{
name: "mcp-playwright-server",
version: "1.0.0",
},
{
capabilities: {
tools: {},
},
}
);5. Creating Tools
Tool: Open a Web Page
server.tool(
"open_page",
{
url: { type: "string" },
},
async ({ url }) => {
const browser = await chromium.launch({ headless: true });
const page = await browser.newPage();
await page.goto(url);
const title = await page.title();
await browser.close();
return {
content: [{ type: "text", text: `Page title: ${title}` }],
};
}
);Tool: Take Screenshot
server.tool(
"take_screenshot",
{
url: { type: "string" },
},
async ({ url }) => {
const browser = await chromium.launch({ headless: true });
const page = await browser.newPage();
await page.goto(url);
await page.screenshot({ path: "screenshot.png" });
await browser.close();
return {
content: [{ type: "text", text: "Screenshot saved successfully" }],
};
}
);6. Running the Server
const transport = new StdioServerTransport();
await server.connect(transport);
// Run with:
// node index.jsConclusion
MCP + Playwright is a game-changer for AI-driven automation. By giving LLMs direct access to the web, we move from simple text generation to actual task execution. Secure, structured, and powerful—this is the future of AI agents.