An MCP (Model Context Protocol) server for penetration testing tools, designed to work with various LLM clients like Claude Desktop, Roo Code, and other compatible MCP clients.
Features
- Comprehensive pentesting tools:
- Directory scanning (FFuf, Dirsearch)
- Vulnerability scanning (Nuclei, XSStrike)
- API testing
- Reconnaissance
- And more...
- Pre-configured wordlists from SecLists
- Automated report generation
- Claude Desktop integration
Prerequisites
- Docker and Docker Compose (for containerized setup)
- Claude Desktop application or other MCP-compatible client
- Python 3.10+ and uv (for local setup)
Directory Setup
- Create the required directories:
# Create directories
mkdir -p reports templates wordlists
- Directory structure should look like this:
pentest-tools/
├── reports/ # For storing scan reports
├── templates/ # For report templates
├── wordlists/ # For custom wordlists
├── pentest-tools-mcp-server.py
├── config.json
├── requirements.txt
├── docker-compose.yml
└── Dockerfile
Setup
Docker Setup (Recommended)
- Build and start the container:
docker-compose up -d --build
- Verify the container is running:
- Check logs if needed:
Local Setup
- Install dependencies:
uv venv
source .venv/bin/activate # On Windows: .venv\Scripts\activate
uv pip install -r requirements.txt
- Install required system tools (example for Ubuntu/Debian):
sudo apt-get install nmap whatweb dnsrecon theharvester ffuf dirsearch sqlmap
Claude Desktop Integration
- Configure Claude Desktop:
Windows:
%APPDATA%\Claude\claude_desktop_config.json
MacOS/Linux:
~/Library/Application Support/Claude/claude_desktop_config.json
- Add server configuration:
For Docker setup:
{
"mcpServers": {
"pentest-tools": {
"command": "docker-compose",
"args": [
"run",
"--rm",
"pentest-tools",
"python3",
"pentest-tools-mcp-server.py"
],
"cwd": "\\Path\\to\\pentest-tools"
}
}
}
If the above configuration doesn't work on Windows, try this alternative approach:
{
"mcpServers": {
"pentest-tools": {
"command": "cmd",
"args": [
"/c",
"cd /d \\path\\to\\pentest-tools && docker-compose run --rm pentest-tools python3 pentest-tools-mcp-server.py"
]
}
}
}
Note about cwd (Current Working Directory):
cwd tells Claude Desktop which directory to run the command from- It must be the absolute path to the directory containing
docker-compose.yml - On Windows, use double backslashes (
\\) in paths - On Linux/MacOS, use forward slashes (
/)
- Restart Claude Desktop
Usage
Available commands in Claude Desktop:
- Reconnaissance:
- Directory scanning:
/scan example.com --type directory
- Vulnerability scanning:
/scan example.com --type full
/scan example.com --type xss
/scan example.com --type sqli
/scan example.com --type ssrf
- API testing:
/scan api.example.com --type api
Natural language commands:
- "Run a full security scan on example.com"
- "Check for XSS vulnerabilities on example.com"
- "Perform reconnaissance on example.com"
Directory Structure Details
pentest-tools/
├── reports/ # Scan reports directory
│ ├── recon/ # Reconnaissance reports
│ ├── vulns/ # Vulnerability scan reports
│ └── api/ # API testing reports
├── templates/ # Report templates
│ ├── recon.html # Template for recon reports
│ ├── vuln.html # Template for vulnerability reports
│ └── api.html # Template for API test reports
├── wordlists/ # Custom wordlists
│ ├── SecLists/ # Cloned from SecLists repo
│ ├── custom/ # Your custom wordlists
│ └── generated/ # Tool-generated wordlists
├── pentest-tools-mcp-server.py # Main MCP server
├── config.json # Tool configuration
├── requirements.txt # Python dependencies
├── docker-compose.yml # Docker configuration
└── Dockerfile # Container definition
Security Notes
- Always ensure you have permission to scan targets
- Keep tools and dependencies updated
- Review scan results carefully
- Follow responsible disclosure practices