How do I use Playwright MCP?

1. Click on "Install Server". 2. Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state. 3. In the chat, type @ followed by the MCP server name and your instructions, e.g., "@Playwright MCP Go to news.ycombinator.com and summarize the top 5 headlines." That's it! The server will respond to your query, and you can continue using it as needed. Here is a step-by-step guide with screenshots.

Playwright MCP

FastMCP server with Playwright for browser automation and web scraping. Exposes tools over HTTP for use with MCP clients (e.g. GPT-based agents).

Features

HTTP transport – Server runs as an HTTP service; clients connect via URL.
Long-lived browser – One Chromium instance started at server startup (Option A).
Structured logging – Structlog with configurable level and JSON output.
Modular layout – Config, logging, browser, tools, and server are separate; state is injected via lifespan context.

Tools

Tool	Description
`browser_navigate`	Navigate to a URL
`browser_snapshot`	Get HTML/text snapshot of the current page
`browser_tabs`	List open tabs (pages)
`browser_click`	Click an element by CSS selector
`browser_type`	Type text into an element
`browser_fill_form`	Fill multiple form fields (selector → value map)
`browser_scrape`	Extract content (text, HTML, or links) from current page or URL
`browser_screenshot`	Take a PNG screenshot (base64) of page or element

Requirements

Python 3.10+
Playwright browsers (install after pip install)

Setup

# From project root
pip install -e .
python -m playwright install chromium

Optional: copy .env.example to .env and adjust MCP_HOST, MCP_PORT, HEADLESS, LOG_LEVEL, LOG_JSON.

Run

# After pip install -e .
playwright-mcp

# Or
python -m playwright_mcp.main

Server listens at http://127.0.0.1:8000 by default. MCP endpoint is typically at http://127.0.0.1:8000/mcp (see FastMCP HTTP deployment docs).

Project layout

src/playwright_mcp/
├── main.py           # Entrypoint
├── config/           # Env-based settings
├── logging_/          # Structlog setup and get_logger
├── browser/          # Lifecycle and BrowserState
├── tools/            # Navigation, interaction, scraping, screenshot
└── server/           # FastMCP app and dependency registration

Configuration

Env var	Default	Description
`MCP_HOST`	`127.0.0.1`	HTTP bind host
`MCP_PORT`	`8000`	HTTP bind port
`HEADLESS`	`true`	Run browser headless
`LOG_LEVEL`	`INFO`	Log level (DEBUG, INFO, WARNING, ERROR)
`LOG_JSON`	`false`	Output logs as JSON