microsoft-playwright-cli

Overview

Playwright CLI is a command-line interface for Playwright browser automation, designed as the preferred approach for coding agents over MCP. It provides token-efficient browser control through concise CLI commands rather than loading large tool schemas and accessibility trees into model context, making it better suited for AI agents that need to balance browser automation with large codebases within limited context windows.

Key Features

• Token-efficient design — Unlike MCP-based approaches, the CLI does not force page data into the LLM context, keeping token usage low while maintaining full browser control capabilities.
• Ref-based element targeting — Uses accessibility snapshot refs (e.g., e21, e35) for precise element interaction, allowing agents to click, type, and check elements without fragile CSS selectors.
• Plugin or standalone operation — Can be installed as a Claude Code plugin via the marketplace, or used standalone where agents read capabilities from playwright-cli --help on their own.
• Headed mode support — Pass --headed to the open command to see the browser visually, useful for debugging and development.
• Standard Playwright commands — Supports opening URLs, typing, pressing keys, clicking, checking elements, taking screenshots, and more through familiar Playwright patterns.

Who is this for?

Playwright CLI is for developers and AI coding agents that need browser automation as part of their workflow. It is particularly well-suited for high-throughput coding agents using Claude Code, GitHub Copilot, or similar tools, where keeping token costs low during browser interaction is important.