xss-mcp-tester

For an explanation and POC of what this mcp server does, please visit my article on medium : Mypost. But globally, it's an MCP server for performing XSS tests with AI.

Installation (for Vscode but overall it's the same thing)

Prerequisites

• Python 3.8+
• uv package manager

Install uv (if not already installed)

curl -LsSf https://astral.sh/uv/install.sh | sh

Setup

1. Clone the repository

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

2. Initialize the project

# Initialize uv project
uv init

# Install dependencies
uv add mcp playwright fastmcp

# Install Playwright browsers
uv run playwright install chromium

3. Verify installation

uv pip list

Configuration

Add the following to your MCP client configuration file:

{
  "mcpServers": {
    "XSS tester": {
      "command": "uv",
      "args": [
        "run",
        "--with",
        "mcp[cli]",
        "--with",
        "playwright", 
        "mcp",
        "run",
        "/path/to/your/project/main.py"
      ]
    }
  }
}

Testing

Start the MCP server in VSCode:

Press Ctrl+Shift+P (or Cmd+Shift+P on Mac) Type "MCP: List Server" and select it Choose your XSS tester server from the list and run it

=> Go to chat, set AI as agent, and let's go

Available Tools

🔍 test_xss_url

Tests a URL for XSS vulnerabilities by checking if JavaScript execution is triggered.

• Input: URL with XSS payloads (e.g., https://example.com/search?q=<script>alert(1)</script>)
• Output: Detects if any JavaScript execution occurs (alerts, console.log, document.write, etc.)

📄 get_raw_html_response

Fetches the raw HTML response from a URL using urllib.

• Input: URL and optional max_length parameter
• Output: Raw HTTP response body as sent by the server
• Features: Truncation control, full response option

🔎 search_in_html_response

Searches for specific content within HTML responses.

• Input: URL and search term
• Output: Search results with context (3 lines before/after matches)
• Use case: Finding specific strings, tokens, or patterns in responses

📜 get_javascript_file

Fetches JavaScript files for manual vulnerability analysis.

• Input: JavaScript file URL
• Output: JavaScript source code for AI analysis
• Use case: Analyzing JS files for DOM XSS vulnerabilities

🔧 get_http_headers

Retrieves HTTP response headers for security analysis.

• Input: URL
• Output: All HTTP response headers
• Use case: Analyzing security protections (CSP, X-Frame-Options, etc.)

🎯 make_custom_http_request

Makes HTTP requests with custom headers (authentication, cookies, sessions).

• Input: URL, method, custom headers string, max_length
• Headers format: "Header1: Value1\nHeader2: Value2\nCookie: session=abc123"
• Output: Full HTTP response with request/response headers
• Use case: Testing authenticated endpoints, session-based vulnerabilities