Playwright MCP
✓ Text source-checked Oct 22, 2026A Model Context Protocol server that provides browser automation capabilities using Playwright.
Source-checked against the vendor's site · rebuilt from the vendor's own pagesProfile updated Link checked How we rate
In short
Playwright MCP is a free server enabling LLMs to automate browsers via structured accessibility snapshots. It supports Chrome, Firefox, and more without needing vision models.
Playwright MCP is a Model Context Protocol server that provides browser automation capabilities using Playwright. The server enables LLMs to interact with web pages through structured accessibility snapshots, bypassing the need for screenshots or visually-tuned models.
Playwright MCP is fast and lightweight, using Playwright's accessibility tree rather than pixel-based input. The tool is LLM-friendly, operating purely on structured data without requiring vision models. Playwright MCP provides deterministic tool application, avoiding ambiguity common with screenshot-based approaches.
The server is compatible with VS Code, Cursor, Windsurf, Claude Desktop, Goose, Grok, Junie, or any other MCP client. Playwright MCP requires Node.js 18 or newer to run.
The tool supports browser channels including Chrome, Firefox, WebKit, and MS Edge. A Chrome extension allows users to connect to existing browser tabs to leverage logged-in sessions and browser state.
The Docker implementation currently supports only headless Chromium. A persistent profile can only be used by one browser instance at a time, which causes conflicts for concurrent MCP clients sharing the same workspace.
According to Playwright MCP: Playwright MCP is a Model Context Protocol server that provides browser automation capabilities using Playwright. Its pages list this strength: Playwright MCP is fast and lightweight, Using Playwright's accessibility tree instead of pixel-based input. They also state this limitation: The Docker implementation only supports headless Chromium.
— The AI Tool Bible editorial team
Pros
- ✅ Playwright MCP is fast and lightweight, using Playwright's accessibility tree instead of pixel-based input.
- ✅ The tool provides deterministic application, avoiding ambiguity common with screenshot-based approaches.
- ✅ Playwright MCP supports multiple browser channels including Chrome, Firefox, WebKit, and MS Edge.
- ✅ A Chrome extension allows connecting to existing browser tabs to leverage logged-in sessions and browser state.
Cons
- ⚠️ The Docker implementation only supports headless Chromium.
- ⚠️ A persistent profile can only be used by one browser instance at a time, causing conflicts for concurrent MCP clients sharing the same workspace.
Facts
- Platforms
- chrome-extension (not source-checked)
- API available
- Yes (not source-checked)
- Open source
- Yes (license: Apache-2.0)
- GitHub stars
- 38,778 (checked Oct 29, 2026) (not source-checked)
- Last GitHub push
- Oct 22, 2026 (not source-checked)
- First commit
- Mar 2025 (not source-checked)
Where this tool sits
Use cases
Frequently asked
- How much does Playwright MCP cost?
- Playwright MCP is free. The provided documentation does not list any specific pricing details or subscription fees for using the server.
- Which browsers are supported by Playwright MCP?
- The tool supports Chrome, Firefox, WebKit, and MS Edge. A Chrome extension allows connecting to existing tabs for logged-in sessions. Docker implementation currently supports only headless Chromium.
- What are the system requirements to run it?
- Playwright MCP requires Node.js 18 or newer to run. It is compatible with VS Code, Cursor, Windsurf, Claude Desktop, Goose, Grok, Junie, or any other MCP client.
- How does it differ from screenshot-based automation?
- It uses Playwright's accessibility tree rather than pixel-based input. This makes it fast, lightweight, and LLM-friendly, operating purely on structured data without requiring vision models or visually-tuned models.
- Can multiple clients use the same browser profile?
- No. A persistent profile can only be used by one browser instance at a time. This causes conflicts for concurrent MCP clients sharing the same workspace.