E2e MCP

🚀 End-to-End Testing for MCP Server

https://github.com/user-attachments/assets/b801f239-dc66-4b3b-bcf2-42e2a9a68721

A Model Context Protocol (MCP) server powered by Playwright, which provides automated end-to-end testing and separates testing concerns from MCP clients through dedicated LLM-driven test validation.

⚠️ Important Note

This MCP is a fork of Microsoft's Playwright MCP. We have optimized the Playwright MCP for automated end-to-end testing.

✨ Features

• Fast and lightweight: Utilizes Playwright's accessibility tree instead of pixel input.
• LLM-friendly: Operates entirely on structured data without the need for visual models.
• Deterministic tool application: Avoids the ambiguities common in screenshot methods.

📚 Documentation

Use Cases

• LLM-driven automated testing

Example Configuration

After cloning this repository, build it and add its E2E MCP server to your MCP client as follows: Note that running this MCP server for end-to-end testing requires an OpenAI API key.

npm install
npx playwright install
npm run build

Then:

{
    "mcpServers": {
        "e2e": {
            "command": "node",
            "args": [
                "/Users/Documents/projects/e2e-mcp/lib/program.js",
                "--endtoend",
                "--api-key=<your openai api key>"
            ]
        }
    }
}

User Data Directory

The E2E MCP will launch the Chrome browser with a new profile located at the following positions:

- Windows: `%USERPROFILE%\AppData\Local\ms-playwright\mcp-chrome-profile`
- macOS: `~/Library/Caches/ms-playwright/mcp-chrome-profile`
- Linux: `~/.cache/ms-playwright/mcp-chrome-profile`

All login information will be stored in this profile, and it can be deleted between sessions if you need to clear the offline state.