CalvinBackup/NeuroSploit

Fork 0

mirror of https://github.com/CyberSecurityUP/NeuroSploit.git synced 2026-02-12 14:02:45 +00:00

Files

Joas A Santos 4b9b0d22be Add files via upload

2026-01-14 15:58:19 -03:00

20 KiB

Raw Blame History

NeuroSploit v2

AI-Powered Penetration Testing Framework with Adaptive Intelligence

NeuroSploit v2 is an advanced security assessment framework that combines reconnaissance tools with adaptive AI analysis. It intelligently collects data, analyzes attack surfaces, and performs targeted security testing using LLM-powered decision making.

What's New in v2

Adaptive AI Mode - AI automatically determines if context is sufficient; runs tools only when needed
3 Execution Modes - CLI, Interactive, and guided Experience/Wizard mode
Consolidated Recon - All reconnaissance outputs merged into a single context file
Context-Based Analysis - Analyze pre-collected recon data without re-running tools
Professional Reports - Auto-generated HTML reports with charts and findings

Features
Installation
Quick Start
3 Execution Modes
Workflow
Adaptive AI Mode
Configuration
CLI Reference
Agent Roles
Built-in Tools
Output Files
Examples
Architecture
Security Notice

Features

Core Capabilities

Feature	Description
Adaptive AI	Automatically runs tools when context is insufficient
Multi-Mode	CLI, Interactive, and Wizard execution modes
Consolidated Recon	All tool outputs merged into single context file
Multi-LLM Support	Claude, OpenAI, Gemini, Ollama, LM Studio
Professional Reports	HTML reports with charts and findings
Extensible	Custom agents, tools, and prompts

Security Testing

Category	Tests
Injection	SQL Injection, XSS, Command Injection, Template Injection
File Attacks	LFI, Path Traversal, File Upload, XXE
Server-Side	SSRF, RCE, Deserialization
Authentication	Auth Bypass, IDOR, Session Issues, JWT
Reconnaissance	Subdomain Enum, Port Scan, Tech Detection, URL Collection

Reconnaissance Tools

Tool	Purpose
subfinder, amass, assetfinder	Subdomain enumeration
httpx, httprobe	HTTP probing
gau, waybackurls, waymore	URL collection
katana, gospider	Web crawling
naabu, nmap	Port scanning
nuclei	Vulnerability scanning

Installation

Prerequisites

# Python 3.8+
python3 --version

# Install dependencies
pip3 install -r requirements.txt

Setup

# Clone repository
git clone https://github.com/your-org/NeuroSploitv2.git
cd NeuroSploitv2

# Create config from example
cp config/config-example.json config/config.json

# Edit with your LLM API keys
nano config/config.json

# Create required directories
mkdir -p results reports logs

# Install security tools (recommended)
python3 neurosploit.py --install-tools

Environment Variables

# Set in .bashrc, .zshrc, or .env
export ANTHROPIC_API_KEY="your_key"
export OPENAI_API_KEY="your_key"
export GEMINI_API_KEY="your_key"

Quick Start

Option 1: Wizard Mode (Recommended for beginners)

python3 neurosploit.py -e

Follow the guided prompts to configure your scan.

Option 2: Two-Step Workflow

# Step 1: Run reconnaissance
python3 neurosploit.py --recon example.com

# Step 2: AI analysis
python3 neurosploit.py --input "Find XSS and SQLi vulnerabilities" \
  -cf results/context_*.json \
  --llm-profile claude_opus_default

Option 3: Interactive Mode

python3 neurosploit.py -i

3 Execution Modes

1. CLI Mode

Direct command-line execution with all parameters:

# Reconnaissance
python3 neurosploit.py --recon example.com

# AI Analysis with context
python3 neurosploit.py --input "Analyze for XSS and SQLi" \
  -cf results/context_X.json \
  --llm-profile claude_opus_default

# Full pentest scan
python3 neurosploit.py --scan https://example.com

# Quick scan
python3 neurosploit.py --quick-scan https://example.com

2. Interactive Mode (`-i`)

REPL interface with tab completion:

python3 neurosploit.py -i

        ╔═══════════════════════════════════════════════════════════╗
        ║         NeuroSploitv2 - AI Offensive Security             ║
        ║                  Interactive Mode                         ║
        ╚═══════════════════════════════════════════════════════════╝

NeuroSploit> help
NeuroSploit> recon example.com
NeuroSploit> analyze results/context_X.json
NeuroSploit> scan https://example.com
NeuroSploit> experience
NeuroSploit> exit

Available Commands:

Command	Description
`recon <target>`	Run full reconnaissance
`analyze <file.json>`	LLM analysis of context file
`scan <target>`	Full pentest with tools
`quick_scan <target>`	Fast essential checks
`experience` / `wizard`	Start guided setup
`set_agent <name>`	Set default agent role
`set_profile <name>`	Set LLM profile
`list_roles`	Show available agents
`list_profiles`	Show LLM profiles
`check_tools`	Check installed tools
`install_tools`	Install required tools
`discover_ollama`	Find local Ollama models

3. Experience/Wizard Mode (`-e`)

Guided step-by-step configuration:

python3 neurosploit.py -e

        ╔═══════════════════════════════════════════════════════════╗
        ║       NEUROSPLOIT - EXPERIENCE MODE (WIZARD)              ║
        ║           Step-by-step Configuration                      ║
        ╚═══════════════════════════════════════════════════════════╝

[STEP 1/6] Choose Operation Mode
--------------------------------------------------
  1. AI Analysis   - Analyze recon context with LLM (no tools)
  2. Full Scan     - Run real pentest tools + AI analysis
  3. Quick Scan    - Fast essential checks + AI analysis
  4. Recon Only    - Run reconnaissance tools, save context

[STEP 2/6] Set Target
[STEP 3/6] Context File
[STEP 4/6] LLM Profile
[STEP 5/6] Agent Role
[STEP 6/6] Custom Prompt

============================================================
  CONFIGURATION SUMMARY
============================================================
  Mode:         analysis
  Target:       example.com
  Context File: results/context_20240115.json
  LLM Profile:  claude_opus_default
  Agent Role:   bug_bounty_hunter
  Prompt:       Find XSS and SQLi vulnerabilities...
============================================================

  Execute with this configuration? [Y/n]:

Workflow

Recommended Workflow

┌─────────────────┐     ┌─────────────────┐     ┌─────────────────┐
│   STEP 1        │     │   STEP 2        │     │   STEP 3        │
│   RECON         │────▶│   AI ANALYSIS   │────▶│   REPORT        │
│                 │     │                 │     │                 │
│ - Subdomains    │     │ - Adaptive AI   │     │ - HTML Report   │
│ - URLs          │     │ - Auto-test     │     │ - JSON Results  │
│ - Ports         │     │ - if needed     │     │ - Findings      │
│ - Technologies  │     │                 │     │                 │
└─────────────────┘     └─────────────────┘     └─────────────────┘

Step 1: Reconnaissance

python3 neurosploit.py --recon example.com

Runs all discovery tools and consolidates output:

Subdomain Enumeration: subfinder, amass, assetfinder
HTTP Probing: httpx, httprobe
URL Collection: gau, waybackurls, waymore
Web Crawling: katana, gospider
Port Scanning: naabu, nmap
Vulnerability Scanning: nuclei

Output: results/context_YYYYMMDD_HHMMSS.json

Step 2: AI Analysis

python3 neurosploit.py --input "Test for SQL injection and XSS" \
  -cf results/context_X.json \
  --llm-profile claude_opus_default

The Adaptive AI:

Analyzes your request
Checks if context has sufficient data
Runs additional tests if needed
Provides comprehensive analysis

Adaptive AI Mode

The AI automatically determines if context data is sufficient:

======================================================================
  NEUROSPLOIT ADAPTIVE AI - BUG_BOUNTY_HUNTER
======================================================================
  Mode: Adaptive (LLM + Tools when needed)
  Target: testphp.vulnweb.com
  Context loaded with:
    - Subdomains: 1
    - URLs: 12085
    - URLs with params: 10989
======================================================================

[PHASE 1] Analyzing Context Sufficiency
--------------------------------------------------
  [*] User wants: xss, sqli
  [*] Data sufficient: No
  [*] Missing: XSS test results, SQL injection evidence

[PHASE 2] Collecting Missing Data
--------------------------------------------------
  [!] Context insufficient for: XSS test results
  [*] Running tools to collect data...

  [XSS] Running XSS tests...
  [>] curl: -s -k "http://target.com/search?q=%3Cscript%3Ealert(1)%3C/script%3E"
  [!] FOUND: XSS

  [SQLi] Running SQL Injection tests...
  [>] curl: -s -k "http://target.com/product?id=1'"
  [!] FOUND: SQL Injection

  [+] Ran 15 tool commands to fill context gaps

[PHASE 3] AI Analysis
--------------------------------------------------
  [*] Generating final analysis with collected evidence...
  [+] Analysis complete

How It Works

Scenario	AI Action
Context has XSS evidence	LLM-only analysis (no tools)
Context missing XSS evidence	Run XSS tests, then analyze
User asks for port scan	Check context, run nmap if missing
General analysis request	Use available context data

Supported Auto-Tests

When context is insufficient, AI can automatically run:

Test	Trigger Keywords
XSS	xss, cross-site, reflected, stored
SQLi	sqli, sql, injection, database
LFI	lfi, file, inclusion, traversal
SSRF	ssrf, server-side, request
RCE	rce, command, execution, shell
Crawl	crawl, discover, spider, urls
Port Scan	port, scan, nmap, service

Configuration

config/config.json

{
  "llm": {
    "default_profile": "claude_opus_default",
    "profiles": {
      "claude_opus_default": {
        "provider": "claude",
        "model": "claude-sonnet-4-20250514",
        "api_key": "${ANTHROPIC_API_KEY}",
        "temperature": 0.7,
        "max_tokens": 8192,
        "guardrails_enabled": true,
        "hallucination_mitigation_strategy": "grounding"
      },
      "ollama_local": {
        "provider": "ollama",
        "model": "llama3:8b",
        "api_key": "",
        "temperature": 0.7
      },
      "gpt_4o": {
        "provider": "gpt",
        "model": "gpt-4o",
        "api_key": "${OPENAI_API_KEY}",
        "temperature": 0.7
      }
    }
  },
  "agent_roles": {
    "bug_bounty_hunter": {
      "enabled": true,
      "description": "Aggressive bug bounty hunting",
      "llm_profile": "claude_opus_default",
      "tools_allowed": ["subfinder", "nuclei", "sqlmap"]
    },
    "red_team_agent": {
      "enabled": true,
      "description": "Red team operations specialist"
    }
  },
  "tools": {
    "nmap": "/usr/bin/nmap",
    "sqlmap": "/usr/bin/sqlmap",
    "nuclei": "/usr/local/bin/nuclei"
  }
}

LLM Providers

Provider	Config Value	Notes
Claude (Anthropic)	`"provider": "claude"`	Best for security analysis
OpenAI	`"provider": "gpt"`	GPT-4, GPT-4o
Google	`"provider": "gemini"`	Gemini Pro
Ollama	`"provider": "ollama"`	Local models
LM Studio	`"provider": "lmstudio"`	Local with OpenAI API

CLI Reference

usage: neurosploit.py [-h] [--recon TARGET] [--context-file FILE]
                      [--target TARGET] [--scan TARGET] [--quick-scan TARGET]
                      [--install-tools] [--check-tools] [-r AGENT_ROLE] [-i]
                      [-e] [--input INPUT] [--llm-profile LLM_PROFILE]

NeuroSploitv2 - AI-Powered Penetration Testing Framework

Arguments:
  --recon TARGET        Run FULL RECON on target
  --context-file, -cf   Load recon context from JSON file
  --target, -t          Specify target URL/domain
  --scan TARGET         Run FULL pentest scan with tools
  --quick-scan TARGET   Run QUICK pentest scan
  --install-tools       Install required security tools
  --check-tools         Check status of installed tools
  -r, --agent-role      Agent role to execute (optional)
  -i, --interactive     Start interactive mode
  -e, --experience      Start wizard mode (guided setup)
  --input               Input prompt for the AI agent
  --llm-profile         LLM profile to use
  --list-agents         List available agent roles
  --list-profiles       List LLM profiles
  -v, --verbose         Enable verbose output

Agent Roles

Predefined agents in config.json with prompts in prompts/:

Agent	Description
`bug_bounty_hunter`	Web app vulnerabilities, high-impact findings
`red_team_agent`	Simulated attack campaigns
`blue_team_agent`	Threat detection and response
`exploit_expert`	Exploitation strategies and payloads
`pentest_generalist`	Broad penetration testing
`owasp_expert`	OWASP Top 10 assessment
`malware_analyst`	Malware examination and IOCs

Custom Agents

Create prompt file: prompts/my_agent.md
Add to config:

"agent_roles": {
  "my_agent": {
    "enabled": true,
    "description": "My custom agent",
    "llm_profile": "claude_opus_default"
  }
}

Built-in Tools

Reconnaissance

Tool	File	Features
OSINT Collector	`tools/recon/osint_collector.py`	IP resolution, tech detection, email patterns
Subdomain Finder	`tools/recon/subdomain_finder.py`	CT logs, DNS brute-force
DNS Enumerator	`tools/recon/dns_enumerator.py`	A, AAAA, MX, NS, TXT, CNAME
Full Recon Runner	`tools/recon/recon_tools.py`	Orchestrates all recon tools

Post-Exploitation

Tool	File	Features
SMB Lateral	`tools/lateral_movement/smb_lateral.py`	Share enum, pass-the-hash
SSH Lateral	`tools/lateral_movement/ssh_lateral.py`	SSH tunnels, key enum
Cron Persistence	`tools/persistence/cron_persistence.py`	Linux persistence
Registry Persistence	`tools/persistence/registry_persistence.py`	Windows persistence

Output Files

File	Location	Description
Context JSON	`results/context_*.json`	Consolidated recon data
Context TXT	`results/context_*.txt`	Human-readable context
Campaign JSON	`results/campaign_*.json`	Full execution results
HTML Report	`reports/report_*.html`	Professional report with charts

HTML Report Features

Executive summary
Severity statistics with charts
Risk score calculation
Vulnerability details with PoCs
Remediation recommendations
Modern dark theme UI

Examples

Basic Recon

# Domain recon
python3 neurosploit.py --recon example.com

# URL recon
python3 neurosploit.py --recon https://example.com

AI Analysis

# Specific vulnerability analysis
python3 neurosploit.py --input "Find SQL injection and XSS vulnerabilities. Provide PoC with CVSS scores." \
  -cf results/context_20240115.json \
  --llm-profile claude_opus_default

# Comprehensive assessment
python3 neurosploit.py --input "Perform comprehensive security assessment. Analyze attack surface, test for OWASP Top 10, prioritize critical findings." \
  -cf results/context_X.json

Pentest Scan

# Full scan with context
python3 neurosploit.py --scan https://example.com -cf results/context_X.json

# Quick scan
python3 neurosploit.py --quick-scan https://example.com -r bug_bounty_hunter

Wizard Mode

python3 neurosploit.py -e
# Follow interactive prompts...

Architecture

NeuroSploitv2/
├── neurosploit.py              # Main entry point
├── config/
│   ├── config.json             # Configuration
│   └── config-example.json     # Example config
├── agents/
│   └── base_agent.py           # Adaptive AI agent
├── core/
│   ├── llm_manager.py          # LLM provider abstraction
│   ├── context_builder.py      # Recon consolidation
│   ├── pentest_executor.py     # Tool execution
│   ├── report_generator.py     # Report generation
│   └── tool_installer.py       # Tool installation
├── tools/
│   ├── recon/
│   │   ├── recon_tools.py      # Advanced recon
│   │   ├── osint_collector.py  # OSINT gathering
│   │   ├── subdomain_finder.py # Subdomain enum
│   │   └── dns_enumerator.py   # DNS enumeration
│   ├── lateral_movement/
│   │   ├── smb_lateral.py      # SMB techniques
│   │   └── ssh_lateral.py      # SSH techniques
│   └── persistence/
│       ├── cron_persistence.py # Linux persistence
│       └── registry_persistence.py # Windows persistence
├── prompts/
│   ├── library.json            # Prompt library
│   └── *.md                    # Agent prompts
├── results/                    # Output directory
├── reports/                    # Generated reports
└── logs/                       # Log files

Security Features

Secure Tool Execution: shlex parsing, no shell injection
Input Validation: Tool paths and arguments validated
Timeout Protection: 60-second default timeout
Permission System: Agent-based tool access control
Error Handling: Comprehensive logging

Troubleshooting

LLM Connection Issues

# Check API key
echo $ANTHROPIC_API_KEY

# Test with local Ollama
python3 neurosploit.py -i
NeuroSploit> discover_ollama

Missing Tools

# Check status
python3 neurosploit.py --check-tools

# Install
python3 neurosploit.py --install-tools

Permission Issues

mkdir -p results reports logs
chmod 755 results reports logs

Security Notice

This tool is for authorized security testing only.

Only test systems you own or have written permission to test
Follow responsible disclosure practices
Comply with all applicable laws and regulations
Unauthorized access to computer systems is illegal

License

MIT License - See LICENSE for details.

Contributing

Fork the repository
Create a feature branch
Submit a pull request

Acknowledgements

LLM Providers

Anthropic Claude
OpenAI GPT
Google Gemini
Ollama
LM Studio

Security Tools

Nmap, Nuclei, SQLMap
Subfinder, Amass, httpx
Katana, Gospider, gau

NeuroSploit v2 - Intelligent Adaptive Security Testing

20 KiB Raw Blame History

NeuroSploit v2

What's New in v2

Table of Contents

Features

Core Capabilities

Security Testing

Reconnaissance Tools

Installation

Prerequisites

Setup

Environment Variables

Quick Start

Option 1: Wizard Mode (Recommended for beginners)

Option 2: Two-Step Workflow

Option 3: Interactive Mode

3 Execution Modes

1. CLI Mode

2. Interactive Mode (-i)

3. Experience/Wizard Mode (-e)

Workflow

Recommended Workflow

Step 1: Reconnaissance

Step 2: AI Analysis

Adaptive AI Mode

How It Works

Supported Auto-Tests

Configuration

config/config.json

LLM Providers

CLI Reference

Agent Roles

Custom Agents

Built-in Tools

Reconnaissance

Post-Exploitation

Output Files

HTML Report Features

Examples

Basic Recon

AI Analysis

Pentest Scan

Wizard Mode

Architecture

Security Features

Troubleshooting

LLM Connection Issues

Missing Tools

Permission Issues

Security Notice

License

Contributing

Acknowledgements

LLM Providers

Security Tools

20 KiB

Raw Blame History

2. Interactive Mode (`-i`)

3. Experience/Wizard Mode (`-e`)