Playwright MCP Automation Server

This project integrates Playwright with the MCP (Modular Command Protocol) framework to enable web automation and testing through natural language or programmatic interaction. It allows you to:

• Open websites and extract information.
• Click elements and perform actions.
• Take screenshots.
• Generate test cases.
• Create JSON test reports.

You can prompt tools like Claude (or any other LLM) to perform browser automation, generate tests, and provide feedback.

---

Features

• Web Automation: Interact with websites using Playwright.
• Test Case Generation: Create automated test scripts.
• Screenshots: Capture web pages for reporting.
• JSON Reports: Save test outcomes for analysis.

---

Project Structure

playwright-mcp/

• playwright_mcp.py — MCP Server for Playwright automation
• mcp-config.json — MCP Server configuration
• playwright_tests/ — Generated test cases
• reports/ — Screenshots and test reports
• requirements.txt — Required dependencies
• README.md — Project documentation

---

Prerequisites

Make sure you have Python 3.8+ installed.

1. Clone the Repository:

git clone https://github.com/yourusername/playwright-mcp.git
cd playwright-mcp

2. Set Up a Virtual Environment (Optional):

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install Required Dependencies:

pip install -r requirements.txt
playwright install

---

Configuration

mcp-config.json — MCP configuration file.

{
    "mcpServers": {
        "playwright-mcp": {
            "command": "python",
            "args": ["playwright_mcp.py"],
            "host": "127.0.0.1",
            "port": 8080,
            "timeout": 30000
        }
    }
}

---

Usage

1. Run the MCP Server:

python playwright_mcp.py

You should see:

🚀 Starting Playwright MCP server on 127.0.0.1:8080

---

Available Tools

1. Open Website

{
  "tool": "open_website",
  "inputs": {
    "url": "https://example.com"
  }
}

Result:

{
  "result": "Page title: Example Domain"
}

---

2. Click Element

{
  "tool": "click_element",
  "inputs": {
    "url": "https://example.com",
    "selector": "button#submit"
  }
}

Result:

{
  "result": "Clicked element with selector: button#submit"
}

---

3. Extract Text

{
  "tool": "extract_text",
  "inputs": {
    "url": "https://example.com",
    "selector": "h1"
  }
}

Result:

{
  "result": "Extracted text: Example Domain"
}

---

4. Take Screenshot

{
  "tool": "take_screenshot",
  "inputs": {
    "url": "https://example.com",
    "filename": "example_screenshot.png"
  }
}

Result:

{
  "result": "Screenshot saved to reports/example_screenshot.png"
}

---

5. Generate Test Case

{
  "tool": "generate_test_case",
  "inputs": {
    "url": "https://example.com",
    "action": "click",
    "selector": "button#submit"
  }
}

Result:

{
  "result": "Test case saved to playwright_tests/test_123456.py"
}

---

6. Generate Report

{
  "tool": "generate_report",
  "inputs": {
    "status": "Passed",
    "details": "All tests passed successfully."
  }
}

Result:

{
  "result": "Report saved to reports/report_123456.json"
}

---

Stopping the Server

Press CTRL+C to stop the MCP server.

---

Extending the Project

You can create additional tools by adding more @mcp.tool() functions to the playwright_mcp.py file.

For example:

@mcp.tool()
async def get_url(page_url: str) -> str:
    """Returns the current URL."""
    async with async_playwright() as pw:
        browser = await pw.chromium.launch(headless=False)
        page = await browser.new_page()
        await page.goto(page_url)
        current_url = page.url
        await browser.close()
        return current_url

---

Contributing

Contributions are welcome! Feel free to open issues or submit pull requests.

---

Troubleshooting

• Playwright Not Found?
  Run:

  playwright install
  

• Port Already in Use?
  Modify the port in mcp-config.json.

---

Contact

For any issues or suggestions, feel free to reach out!

---