AI Pentest Agent

Autonomous penetration testing agent with LLM-guided reasoning, modular scanning, and professional reporting.

What It Does

• Port Scanning: Multi-threaded TCP port scanner with service detection and banner grabbing
• Web Reconnaissance: HTTP analysis, security header checks, technology detection, path discovery, SSL/TLS analysis
• Vulnerability Scanning: CVE matching against known vulnerable service versions, nuclei integration, exposed service detection
• Exploitation Framework: SSH analysis (version checks, weak algorithms), SMB version detection, web vulnerability correlation
• LLM Planning: Optional AI-guided attack chain generation (falls back to heuristic planner)
• Report Generation: HTML and Markdown reports with CVSS scoring
• Web Dashboard: Real-time campaign monitoring
• REST API: Programmatic access to all features

Quick Start

# Install
pip install -r requirements.txt

# Scan a target
pentest-agent scan 127.0.0.1

# Web reconnaissance
pentest-agent webscan example.com --deep

# Vulnerability scan
pentest-agent vulnscan 127.0.0.1

# Full campaign
pentest-agent attack 127.0.0.1 --output report.json

# Start API server
pentest-api
# Then open http://localhost:9090 in browser

# Generate config template
pentest-agent init

Architecture

┌──────────────┐     ┌──────────────┐     ┌──────────────┐
│  Recon       │────▶│  Planner     │────▶│  Executor    │
│  Module      │     │  (LLM/H)     │     │  (Tools)     │
└──────────────┘     └──────────────┘     └──────┬───────┘
                                                  │
┌──────────────┐     ┌──────────────┐     ┌──────▼───────┐
│  Reporter    │◀────│  Vuln Scan   │◀────│  Exploit     │
│  (HTML/MD)   │     │  (CVE/Nuc)   │     │  Framework   │
└──────────────┘     └──────────────┘     └──────────────┘

Modules

Module	File	Description
Port Scanner	modules/scanner.py	Multi-threaded TCP scanner
Web Recon	modules/web_recon.py	HTTP analysis, headers, paths
Vuln Scanner	modules/vuln_scanner.py	CVE matching, nuclei wrapper
Exploits	modules/exploits.py	SSH, SMB, web exploit modules
Report Gen	modules/report_gen.py	HTML/Markdown reports
Red Team	agents/red_team.py	Campaign orchestrator
Planner	core/planner.py	LLM + heuristic planning
Safety	core/safety.py	Scope validation, blocked targets
Config	core/config.py	YAML config, env vars

Safety

• Scope validation against allowed target lists
• Blocked IP ranges (loopback, link-local, multicast)
• Blocked domains (.gov, .mil, major cloud providers)
• Destructive action approval gates
• Custom script dangerous pattern detection

Requirements

• Python 3.10+
• Optional: nmap, nuclei, sqlmap (for enhanced scanning)
• Optional: OpenAI API key (for LLM-guided planning)

License

Apache-2.0