Skip to content

Open-source framework

REDVECTOR AI

AI-Assisted Penetration Testing & Red Team Orchestration

Coordinate penetration testing and vulnerability assessment through 328 registered MCP tool interfaces and an extensible Python/Flask backend.

Explore on GitHub View Installation Guide Explore Capabilities
  • Open Source
  • MIT License
  • Python / Flask
  • Model Context Protocol
  • Kali Linux
RedVector logo: a red cybernetic dragon inside a shield above the RedVector wordmark
redvector: terminalPreview
 

RedVector Terminal Preview

0
Registered MCP Tool Interfaces
0
Flask API Routes
Python
Execution Backend
MCP
AI Client Integration

328 Registered MCP Tool Interfaces

// capabilities

Built for professional security work

Eight assessment disciplines accessible through a consistent MCP interface.

  • AI-Assisted Security Orchestration

    Coordinate security tools and assessment workflows through MCP-compatible AI clients.

  • Network Reconnaissance

    Integrations for network discovery, port scanning, service enumeration and infrastructure analysis.

  • Vulnerability Assessment

    Identify and investigate vulnerabilities with supported scanners and security intelligence tools.

  • Web & API Security

    Assess web applications and APIs with HTTP inspection, fuzzing and authentication testing.

  • Active Directory Security

    Assess Active Directory and Windows environments through AD and SMB integrations.

  • Cloud & Container Security

    AWS, Azure, GCP, container, Kubernetes and infrastructure-as-code security assessments.

  • Digital Forensics & Reverse Engineering

    Memory analysis, binary inspection, firmware research and digital evidence workflows.

  • MCP AI Client Integration

    Expose registered tool interfaces to MCP-compatible clients, including Claude Desktop.

// architecture

From AI client to structured results

MCP requests pass through the Python/Flask backend to host-installed security tools.

AI Client / MCP-Compatible Agent

An MCP-compatible AI client or agent sends tool requests to RedVector.

RedVector Architecture Overview

// security tools

Security tool integrations

Selected integrations for reconnaissance, vulnerability assessment and security research.

Network Reconnaissance

  • Nmap
  • Masscan
  • Rustscan
  • Amass
  • Subfinder
  • httpx

Web Application Security

  • OWASP ZAP
  • ffuf
  • Nikto
  • SQLMap
  • Gobuster
  • Feroxbuster

Vulnerability Assessment

  • Nuclei
  • Metasploit
  • Searchsploit
  • WPScan

Active Directory & Windows

  • NetExec
  • BloodHound
  • Impacket
  • Responder
  • Hashcat
  • John the Ripper

Cloud & Container Security

  • Prowler
  • Trivy
  • Scout Suite
  • Checkov
  • kube-hunter

Digital Forensics

  • Volatility
  • Binwalk
  • ExifTool
  • Foremost

Reverse Engineering

  • Ghidra
  • Radare2
  • GDB
  • Binwalk

OSINT & Security Intelligence

  • theHarvester
  • Shodan
  • Sherlock
  • Amass

Tool availability depends on installed dependencies and host configuration. See the documentation for integration requirements.

Full tool documentation

// workflow

How RedVector works

A typical workflow from backend initialization to validated findings.

// startup

The command-center view

redvector: startupPreview
 

RedVector Startup Preview

// installation

Get started

Install the Python dependencies and configure your MCP client. Security tools must be installed separately on the host.

Clone & install dependencies · bash
git clone https://github.com/0x7AK7/RedVector.git
cd RedVector
 
python3 -m venv .venv
source .venv/bin/activate
 
python -m pip install --upgrade pip
pip install -r requirements.txt
Standalone MCP operation · bash
bash start_redvector.sh
Backend only · bash
python redvector_server.py

The MCP launcher uses stdio. Once configured, Claude Desktop launches start_redvector.sh automatically. Do not start the launcher separately for that session.

Complete Installation Documentation

// design philosophy

Why RedVector

Architecture and design principles behind the framework.

PRINCIPLE 01

MCP-compatible tool interfaces

Registered handlers expose security tools to AI clients through the Model Context Protocol.

PRINCIPLE 02

Cross-discipline security tooling

Network, web, cloud, Active Directory, forensics and reverse engineering in one framework.

PRINCIPLE 03

Extensible Python backend

An open Python/Flask codebase for custom integrations and assessment workflows.

PRINCIPLE 04

Structured security workflows

Decision, workflow and execution components organize assessments.

PRINCIPLE 05

Process monitoring and reporting

Monitor execution, cache results and collect telemetry for long-running assessments.

PRINCIPLE 06

Kali Linux compatibility

Designed to work with the tooling commonly found on Kali Linux.

PRINCIPLE 07

Open-source development

Source code, issues and releases are public on GitHub under the MIT license.

PRINCIPLE 08

Controlled deployment

Deployment guidance covers authorized testing and keeping the backend off the public internet.

OPEN SOURCE. BUILT FOR SECURITY RESEARCH.

Explore the RedVector source code, review its architecture, follow development and contribute to the project.

GitHub Repository Read Documentation Report an Issue View Releases Star on GitHub

// creator

Built by Adam Karim

Adam Karim develops and maintains RedVector AI for practical penetration testing, digital forensics and AI-assisted security research.

GitHub Profile