Skip to main content
Playwright MCP provides a comprehensive set of tools for browser automation through the Model Context Protocol. These tools enable AI assistants to interact with web pages, manage browser tabs, perform coordinate-based interactions, generate PDFs, and execute test assertions.

Tool Categories

Playwright MCP organizes its tools into the following categories:

Core Automation

Essential tools for browser navigation, element interaction, form filling, and page inspection

Tab Management

Tools for creating, listing, closing, and switching between browser tabs

Coordinate-Based

Low-level mouse operations using screen coordinates (requires vision capability)

PDF Generation

Save web pages as PDF documents (requires pdf capability)

Test Assertions

Verify page elements and content for testing (requires testing capability)

Opt-in Capabilities

Some tool categories require explicit opt-in through the --caps flag when starting the server. This provides fine-grained control over which capabilities are enabled.

Enabling Capabilities

Use the --caps flag to enable specific capabilities:
# Enable vision capability for coordinate-based tools
npx @playwright/mcp-server --caps=vision

# Enable multiple capabilities
npx @playwright/mcp-server --caps=vision,pdf,testing

Available Capabilities

Enables coordinate-based mouse operations using x/y coordinates. Required for tools like browser_mouse_click_xy, browser_mouse_move_xy, and browser_mouse_drag_xy.
Enables PDF generation functionality. Required for the browser_pdf_save tool.
Enables test assertion tools for verifying page elements and content. Required for tools like browser_verify_element_visible and browser_verify_text_visible.
Enables Playwright tracing functionality for debugging and performance analysis.

Read-Only vs. Write Operations

Tools are categorized by whether they modify browser state:
  • Read-only tools: Safe operations that only read information (e.g., browser_snapshot, browser_take_screenshot, browser_console_messages)
  • Write tools: Operations that modify the browser or page state (e.g., browser_click, browser_type, browser_navigate)
Read-only tools are useful for inspection and debugging without affecting the page state.

Next Steps

Core Automation Tools

Start with essential browser automation tools

Quick Start Guide

Set up your first Playwright MCP integration