Lets LLMs interact with web pages using structured accessibility snapshots instead of screenshots. The README also points coding agents to the Playwright CLI with skills as a more token-efficient alternative.
What it does
Where Playwright MCP runs and what it adds, from its documentation
Drive a browser via the accessibility tree
Deterministic tool calls without pixel-based input
Persistent profile, isolated contexts, or connect to an existing browser via extension
Headless or headed mode
Restrict origins with allowed/blocked origin options
When to use it
Good fit
Exploratory automation and self-healing tests
Long-running agent loops that need persistent browser state
Not for
High-throughput coding agents with tight context: README suggests Playwright CLI with skills
Using origin lists as a security boundary: README says they are not one
How to install Playwright MCP
You need: Node.js 18 or newer; An MCP client such as VS Code, Cursor, Windsurf, Claude Desktop
Install Node.js 18 or newer
Claude Code: claude mcp add playwright npx @playwright/mcp@latest