
Playwright MCP
Microsoft's official browser-automation MCP server — let agents drive a real browser with Playwright.
About
microsoft/playwright-mcp (38k stars, Apache-2.0) exposes Playwright's browser automation as MCP tools, so AI agents can navigate pages, click, fill forms, take screenshots, and run web tests in a real Chromium/Firefox/WebKit browser.
Its README documents install recipes for ~20 clients — Claude Code, Antigravity, Cursor, Windsurf, Copilot, Cline, Codex, Gemini CLI, opencode, Amp, and more — which makes it the best-documented compatibility matrix in the MCP space. 39 releases in the last 12 months; pushed Oct 2026.
What it can do
- Full Playwright browser automation (navigate, click, type, screenshots, network) as MCP tools
- ~20 documented client integrations — the widest compatibility matrix of any MCP server checked
- Apache-2.0, Microsoft-maintained, 39 releases in 12 months
- Snapshot-based accessibility tree keeps agent context compact
- Works headless for CI/test pipelines as well as interactive agent use
Setup
- No install needed for the quick path:
npx @playwright/mcp@latest. - Claude Code:
claude mcp add playwright npx @playwright/mcp@latest. - Or add the stdio config block from the README to your client's MCP settings.
- The server launches its own browser — no separate Playwright install required.
- Scope agents to trusted sites; review what the agent is allowed to submit.
Prerequisites
Node.js (for npx); an MCP-compatible client; a display or headless-capable environment.
Use cases
Works with
Trust score — 9/10
Scored 9/10 (verified tier). Microsoft-maintained, 38k stars, Apache-2.0, and the rare MCP server whose README documents per-client setup for ~20 agents including Antigravity and Claude Code. Docked one point: a browser-driving agent is inherently powerful — form submissions, authenticated sessions, and file downloads all need explicit scoping, which the tool leaves to you.
★ 38,067 GitHub stars
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 8.5/10
- Maintenance · recency, releases, issue load
- 10.0/10
- Openness · license + docs
- 10.0/10
- Manual review · sanity read + red-flag check
- 9.0/10
- · 38,067 GitHub stars (log-scaled vs catalog peers)
- · last push 4d ago; 39 releases/12mo; 0 open issues
- · manual review: Microsoft browser-automation server (38k stars), ~20 documented client integrations incl. Antigravity; browser-driving agents need explicit scoping
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.