MseeP.ai Security Assessment	Special thanks to Warp, the AI terminal for developers

A Model Context Protocol server that provides browser automation capabilities using Playwright. This server enables LLMs to interact with web pages, take screenshots, generate test code, web scrapes the page and execute JavaScript in a real browser environment.

Test your web applications on real device profiles with a simple command:

// Test on iPhone 13 with automatic user-agent, touch support, and device pixel ratio
await playwright_resize({ device: "iPhone 13" });

// Switch to iPad with landscape orientation
await playwright_resize({ device: "iPad Pro 11", orientation: "landscape" });

// Test desktop view
await playwright_resize({ device: "Desktop Chrome" });

Natural Language Support for AI Assistants:

• "Test on iPhone 13"
• "Switch to iPad view"
• "Rotate to landscape"

Supports 143 devices: iPhone, iPad, Pixel, Galaxy, and Desktop browsers with proper emulation of viewport, user-agent, touch events, and device pixel ratios.

📚 View Device Quick Reference | Prompt Guide

You can install the package using either npm, mcp-get, or Smithery:

Using npm:

npm install -g @executeautomation/playwright-mcp-server

Using mcp-get:

npx @michaellatman/mcp-get@latest install @executeautomation/playwright-mcp-server

Using Smithery

To install Playwright MCP for Claude Desktop automatically via Smithery:

npx @smithery/cli install @executeautomation/playwright-mcp-server --client claude

Using Claude Code:

claude mcp add --transport stdio playwright npx @executeautomation/playwright-mcp-server

Install the Playwright MCP server in VS Code using one of these buttons:

Alternatively, you can install the Playwright MCP server using the VS Code CLI:

# For VS Code
code --add-mcp '{"name":"playwright","command":"npx","args":["@executeautomation/playwright-mcp-server"]}'

# For VS Code Insiders
code-insiders --add-mcp '{"name":"playwright","command":"npx","args":["@executeautomation/playwright-mcp-server"]}'

After installation, the ExecuteAutomation Playwright MCP server will be available for use with your GitHub Copilot agent in VS Code.

This is the recommended mode for Claude Desktop.

{
  "mcpServers": {
    "playwright": {
      "command": "npx",
      "args": ["-y", "@executeautomation/playwright-mcp-server"]
    }
  }
}

Note: In stdio mode, logging is automatically directed to files only (not console) to maintain clean JSON-RPC communication. Logs are written to ~/playwright-mcp-server.log.

When running headed browser on systems without display or from worker processes of IDEs, you can run the MCP server as a standalone HTTP server:

Note for Claude Desktop Users: Claude Desktop currently requires stdio mode (command/args configuration). HTTP mode is recommended for VS Code, custom clients, and remote deployments. See CLAUDE_DESKTOP_CONFIG.md for details.

# Using npx
npx @executeautomation/playwright-mcp-server --port 8931

# Or after global installation
playwright-mcp-server --port 8931

The server will start and display available endpoints:

==============================================
Playwright MCP Server (HTTP Mode)
==============================================
Port: 8931

ENDPOINTS:
- SSE Stream:     GET  http://localhost:8931/sse
- Messages:       POST http://localhost:8931/messages?sessionId=<id>
- MCP (unified):  GET  http://localhost:8931/mcp
- MCP (unified):  POST http://localhost:8931/mcp?sessionId=<id>
- Health Check:   GET  http://localhost:8931/health
==============================================

⚠️ CRITICAL: The "type": "http" field is REQUIRED for HTTP/SSE transport!

For VS Code GitHub Copilot:

{
  "github.copilot.chat.mcp.servers": {
    "playwright": {
      "url": "http://localhost:8931/mcp",
      "type": "http"
    }
  }
}

For Custom MCP Clients:

{
  "mcpServers": {
    "playwright": {
      "url": "http://localhost:8931/mcp",
      "type": "http"
    }
  }
}

Important: Without "type": "http", the connection will fail.

For Claude Desktop: Use stdio mode instead (see Standard Mode above)

• Running headed browsers on systems without display (e.g., remote servers)
• Integrating with VS Code GitHub Copilot
• Running the server as a background service
• Accessing the server from multiple clients
• Debugging with the /health endpoint
• Custom MCP client integrations

Monitoring: The server includes a monitoring system that starts on a dynamically allocated port (avoiding conflicts). Check the console output for the actual port.

Note: For Claude Desktop, continue using stdio mode (Standard Mode above) for now.

Symptom: 400 error with message "Bad Request: No transport found for sessionId"

Solution:

1. Check configuration includes "type": "http"

   {
     "url": "http://localhost:8931/mcp",
     "type": "http"  // ← This is REQUIRED!
   }

2. Verify server logs show connection:

   # Should see these in order:
   # 1. "Incoming request" - GET /mcp
   # 2. "Transport registered" - with sessionId
   # 3. "POST message received" - with same sessionId

3. Restart both server and client

• Server not starting: Check if port 8931 is available
• External access blocked: This is by design (security). Server binds to localhost only
• For remote access: Use SSH tunneling:

  ssh -L 8931:localhost:8931 user@remote-server

This project uses Jest for testing. The tests are located in the src/__tests__ directory.

You can run the tests using one of the following commands:

# Run tests using the custom script (with coverage)
node run-tests.cjs

# Run tests using npm scripts
npm test           # Run tests without coverage
npm run test:coverage  # Run tests with coverage
npm run test:custom    # Run tests with custom script (same as node run-tests.cjs)

The test coverage report will be generated in the coverage directory.

The evals package loads an mcp client that then runs the index.ts file, so there is no need to rebuild between tests. You can load environment variables by prefixing the npx command. Full documentation can be found here.

OPENAI_API_KEY=your-key  npx mcp-eval src/evals/evals.ts src/tools/codegen/index.ts

When adding new tools, please be mindful of the tool name length. Some clients, like Cursor, have a 60-character limit for the combined server and tool name (server_name:tool_name).

Our server name is playwright-mcp. Please ensure your tool names are short enough to not exceed this limit.