What Is METATRON? AI Pentesting That Works Offline
What Is METATRON? AI Pentesting That Works Offline
METATRON is an open-source AI penetration testing assistant that runs entirely on your local machine. No cloud, no API keys, no subscriptions. Released in early April 2026, it’s gaining rapid attention in the security community for bringing AI-powered vulnerability assessment fully offline.
Last verified: April 12, 2026
How It Works
METATRON is a CLI tool with a simple workflow:
- You provide a target — IP address or domain
- METATRON runs recon tools — nmap, whois, whatweb, curl, dig, nikto
- Local LLM analyzes results — Identifies vulnerabilities, suggests next steps
- Everything stays local — No data leaves your machine
# Example usage
metatron scan 192.168.1.100
metatron scan example.com --full
Key Features
| Feature | Details |
|---|---|
| Fully offline | No internet required after setup |
| Local LLM | Runs via Ollama (Llama, Mistral, etc.) |
| No API costs | Zero ongoing costs |
| Linux-native | Built for Parrot OS, works on most Linux distros |
| CLI-based | Terminal interface, scriptable |
| Open source | GitHub: sooryathejas/METATRON |
| Recon tools | nmap, whois, whatweb, curl, dig, nikto |
Why It Matters
Most AI-powered security tools send your scan data to the cloud. That’s a non-starter for:
- Corporate penetration testers scanning internal networks
- Government security teams with air-gapped environments
- Bug bounty hunters who want to keep their findings private
- Training environments where internet access is limited
METATRON solves this by keeping everything local. The AI analysis runs on your hardware, your data never leaves your machine.
Technical Requirements
| Component | Requirement |
|---|---|
| OS | Linux (Parrot OS recommended) |
| LLM Runtime | Ollama |
| RAM | 8GB minimum, 16GB+ recommended |
| Storage | 10GB+ for LLM models |
| GPU | Optional but recommended |
| Network tools | nmap, whois, whatweb, nikto (auto-installed) |
Limitations
- Linux only — No Windows or macOS support yet
- Local LLM quality — Analysis is limited by your local model’s capability
- No cloud-scale intelligence — Doesn’t pull from threat intelligence databases
- Hardware dependent — Better GPU = faster analysis
- Early stage — Active development, expect rough edges
METATRON vs Cloud Pentesting AI
| Feature | METATRON | Cloud AI Pentesting |
|---|---|---|
| Privacy | ✅ Everything local | ❌ Data sent to cloud |
| Cost | Free | $50-500/month |
| Analysis quality | Good (depends on local model) | Better (frontier models) |
| Threat intelligence | No | Yes |
| Air-gapped environments | ✅ | ❌ |
| Setup effort | Medium (install Ollama + tools) | Low (SaaS) |
Getting Started
# Install Ollama for local LLM
curl -fsSL https://ollama.com/install.sh | sh
ollama pull llama3.2
# Clone METATRON
git clone https://github.com/sooryathejas/METATRON.git
cd METATRON
# Install dependencies and run
pip install -r requirements.txt
python metatron.py scan <target>
The Takeaway
METATRON fills a real gap: AI-powered penetration testing that respects privacy and works offline. It’s not a replacement for professional pentesting tools or cloud-based AI analysis with threat intelligence, but for quick recon and vulnerability assessment without sending data anywhere, it’s the best open-source option available.
Last verified: April 12, 2026