Playwright (Cross-browser) MCP Server
Purpose
The Playwright MCP Server exposes Mobitru’s browser grid capabilities through MCP.
It enables fully automated cross-browser testing directly from AI-powered IDEs, with support for Chromium, Chrome, Firefox, and WebKit (Safari).
Primary Use Cases
- Starting and controlling Playwright browser sessions.
- Navigating to URLs, performing DOM interactions, and taking screenshots.
- Simulating user behavior — clicks, typing, scrolling, navigation.
- Running visual regression or layout validation tests.
- Capturing browser-level artifacts such as network files, console logs, and performance data.
Typical Workflow
| Step | Description |
| 1. Initialize browser | Start a new Playwright session specifying browser type (Chromium, Firefox, etc.). |
| 2. Navigate | Load the target URL or web application. |
| 3. Interact | Perform clicks, inputs, or other scripted interactions. |
| 4. Capture data | Take screenshots, save HAR or console logs. |
| 5. Close session | Gracefully terminate the Playwright instance. |
Example Flow
Notes
- Safari support is provided via the WebKit engine or natively.
- Typically, only one active Playwright session is managed at a time per MCP instance.
- Multiple tabs can be opened inside the same session, but concurrent browser types require separate MCP processes.
- Browser-level sessions are ephemeral and reset automatically after closure.