pentester-mcp

Overview

Pentester-MCP transforms AI assistants into autonomous cybersecurity experts by providing access to over 200 open-source penetration testing tools through the Model Context Protocol. The platform offers two deployment methods: a secure Docker sandbox (recommended) that isolates tool execution from the host system, and local execution for users with existing tool installations. Each tool is equipped with AI-optimized documentation strings, enabling the AI to understand tool arguments, syntax, and requirements autonomously.

Try asking AI

After installing, here are 6 things you can ask your AI assistant:

• nmap

  Network scanning tool to discover hosts and services

  Try:Scan the target network for open portsTry:Perform a SYN scan on 192.168.1.0/24
• masscan

  Fast TCP port scanner for large networks

  Try:Scan a target range for open ports at 10 million packets/secondTry:Perform a TCP scan on 10.0.0.0/8
• sqlmap

  Automatic SQL injection and database takeover tool

  Try:Test the login form for SQL injection vulnerabilitiesTry:Extract data from the database using time-based blind injection
• ffuf

  Fuzzing tool for discovering web content and directories

  Try:Discover hidden directories on the target websiteTry:Fuzz for subdomains with a custom wordlist
• gobuster

  Directory/file and DNS busting tool

  Try:Find hidden directories on the target web serverTry:Enumerate subdomains for the target domain
• dirsearch

  Web path scanner for discovering hidden files and directories

  Try:Scan for common directories on the target websiteTry:Discover backup files and hidden endpoints
• nikto

  Web server scanner that tests for dangerous files/CGIs

  Try:Scan the web server for known vulnerabilitiesTry:Test the target for outdated software versions
• hydra

  Online password cracking tool supporting various protocols

  Try:Brute force SSH login credentialsTry:Perform a dictionary attack on a web login form
• john

  Advanced password cracking tool for various hash types

  Try:Crack the captured password hashes using a dictionary attackTry:Perform a brute-force attack on the NTLM hashes
• hashcat

  Fast password cracking tool supporting multiple attack modes

  Try:Crack the MD5 hash using a rule-based attackTry:Perform a mask attack on the captured password hashes
• nuclei

  Vulnerability scanner with templates for detecting security issues

  Try:Scan the target for known vulnerabilitiesTry:Test for specific CVEs in the web application
• recon-ng

  Web-based reconnaissance framework

  Try:Perform OSINT reconnaissance on the target domainTry:Gather information about the target organization

Note: Tools extracted from the 'The Arsenal' section which explicitly lists 235+ penetration testing tools available in the MCP server.

Comparable tools

Installation

git clone https://github.com/halilkirazkaya/pentester-mcp.git
cd pentester-mcp

docker compose up -d --build

{
  "mcpServers": {
    "pentester_mcp": {
      "command": "docker",
      "args": [
        "exec",
        "-i",
        "pentester-mcp",
        "/app/.venv/bin/python",
        "/app/server.py"
      ]
    }
  }
}

git clone https://github.com/halilkirazkaya/pentester-mcp.git
cd pentester-mcp
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

FAQ

Is it safe to run penetration testing tools through AI assistants?

Yes, Pentester-MCP uses a Docker sandbox that isolates tool execution from your host system, preventing any potential harm to your main machine. All tools run within a contained environment.

How many tools are included in Pentester-MCP?

The project provides access to over 200 penetration testing tools across various categories including reconnaissance, web exploitation, active directory, network analysis, and more. New tools are regularly added based on community requests.

Can I add custom tools to Pentester-MCP?

Currently, tools are auto-generated from YAML cheat sheets to ensure consistent API design and security practices. While you can't directly add arbitrary tools, you can request new tools by opening an issue on the GitHub repository.

Compare pentester-mcp with