NullSec Red Team AI

NullSec Red Team AI is an advanced offensive security toolkit designed to integrate seamlessly with Claude Desktop via the Model Context Protocol (MCP). It provides a unified interface for over 150+ professional penetration testing tools and a controlled environment for AI-driven security testing.

Project Overview

This project bridges the gap between Large Language Models and professional security tooling. By providing Claude with direct, controlled access to a specialized security environment, it enables collaborative workflows for reconnaissance, exploitation simulation, and model safety testing.

Integrated Capabilities

Component	Description
HexStrike AI	A Flask-based orchestration server managing 150+ offensive tools (Nmap, SQLMap, etc.).
AI Security Lab	A specialized environment for testing LLM jailbreaks and scanning for model vulnerabilities.
Guardian Tool	A diagnostic utility for system integrity checks and guided repair of the installation.
MCP Bridge	Native integration for Claude Desktop to execute commands and manage a local workspace.

Installation & Deployment

The installer is now unified and hardened, providing a single entry point for a full, secure deployment.

Prerequisites

• Operating System: Kali Linux, Debian, or Ubuntu (recommended).
• Privileges: Sudo access is required for system-wide tool installation.
• Dependencies: Python 3.10+, Node.js 18+, and Git.

Automated Installation

Run the installer with the desired mode. By default, it installs core security tools.

# Clone the repository
git clone https://github.com/Panda1847/NullSec-RedTeam-AI.git
cd NullSec-RedTeam-AI

# Run a dry-run to see planned changes
sudo ./install.sh --dry-run --full

# Perform a full installation (Core + Desktop + MCP + Lab)
sudo ./install.sh

Installation Modes

Mode	Flag	Description
Core	--core	Installs only the 150+ security tools (apt/pip).
Desktop	--desktop	Installs Claude Desktop for Linux.
MCP	--mcp	Configures HexStrike MCP servers in Claude.
Lab	--lab	Deploys the AI Security Lab payloads.
Full	--full	Complete deployment of all components.

Security & Trust

We prioritize transparency and security in our deployment process.

Access & Permissions

• Workspace Isolation: By default, Claude is restricted to the ~/NullSec_RedTeam_Lab directory.
• Elevated Access: Full system access is opt-in only via the --elevated flag during installation.
• Log Privacy: Installation logs are stored at /var/log/nullsec/install.log with restricted permissions (600).
• Reproducibility: All MCP packages are pinned to specific versions to ensure consistent behavior.

Guardian Diagnostic Tool

The guardian tool is deployed to /usr/local/bin/guardian. It is designed to be read-only by default.

• Check Integrity: guardian --check
• Diagnose Issues: guardian "error message"
• Guided Repair: guardian --repair "error message" (Requires explicit user confirmation)