# microsoft/playwright-mcp

Playwright MCP server

Repository: https://github.com/microsoft/playwright-mcp
Canonical: https://ross.abutalabs.com/products/playwright-mcp
Homepage: https://www.npmjs.com/package/@playwright/mcp
Language: TypeScript
License: Apache-2.0
License Family: permissive
Topics: mcp, playwright
Last push: 2026-08-21T22:46:04+00:00

## Health v2 (maintenance only)
Score: 81/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 98, release rhythm 84, longevity 37
- inputs: {"age_days": 530, "days_push": 12, "days_rel": 28, "gap_med": 5.5, "n_releases_24m": 69}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 36504, forks 3065 (observed 2026-08-28T04:12:01.135468+00:00)

## What it is
A Model Context Protocol (MCP) server from Microsoft that exposes Playwright browser automation capabilities to LLMs and MCP clients. It lets language models interact with web pages through structured accessibility snapshots rather than screenshots, enabling deterministic, vision-free browser control.

## Use cases
- let an LLM agent navigate and interact with web pages
- automate browser tasks from Claude Desktop or VS Code via MCP
- build self-healing end-to-end tests driven by an AI agent
- perform exploratory web automation without vision models
- scrape or fill forms on websites through structured accessibility data
- run long-running autonomous browser workflows with persistent state

## When to choose
- you want an MCP client like Claude Desktop, Cursor, or VS Code to control a browser
- you need token-efficient, deterministic browser automation based on accessibility trees instead of screenshots
- you are building agentic loops that benefit from persistent browser state and rich page introspection
- you want a Microsoft-maintained, actively released integration between Playwright and LLMs

## When to avoid
- you are using a high-throughput coding agent where the Playwright CLI with SKILLS is more token-efficient
- you need pixel-level visual interaction that requires screenshot-based vision models
- your project does not use Node.js 18+ or any MCP-compatible client
- you need simple scripted browser automation without any LLM involvement - plain Playwright is a better fit

## Facets
- artifact type: service
- maturity: active
- function: mcp, browser-extension, testing, agent-framework, workflow-automation
- domain: developer-tools, testing, artificial-intelligence
- platform: cross-platform, cli
- tags: playwright, browser-automation, mcp-server, llm-tools, accessibility-tree, web-interaction, ai-agents, automation, nodejs

## Member repositories
- microsoft/playwright-mcp (main) score 81

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:12:01.135468+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T16:26:10.908010+00:00, confidence not recorded.
  - readme: https://github.com/microsoft/playwright-mcp (fetched 2026-08-28T04:12:01.135468+00:00, sha 39c49543584b)
- Data as of 2026-08-30T08:39:29.467469+00:00.