phantomred — scan
$ phantomred --mode full --target scanme.nmap.org
→ Scope check... authorized
→ Consent gate... YES
→ Recon pipeline... running
→ Nmap scan... 2 ports
→ Nuclei... CVE-2023-48795
→ FFUF... /.htaccess [403]
→ AI analysis... dispatching
→ Report generated
 
6 findings in 4m 12s
AI-Powered Security Research

PhRed
Sec

Autonomous Security Intelligence — Est. 2026

Building the next generation of AI-powered security tools for bug bounty hunters and freelance pentesters. Two products. One mission — make professional-grade security accessible to everyone.

PhantomRed → DocSentinel → Research Paper →
6
Findings in 4 minutes
10K+
Nuclei CVE Templates
8B
Llama 3 Parameters
2
Products in Production
Our Products

Two Tools.
One Mission.

Pentesting Platform
PHANTOM
RED
phantomred.com

Autonomous AI penetration testing platform. Chains Llama 3 8B with Nmap, Nuclei, FFUF, and SQLMap into a full end-to-end pipeline with a consent-first ethical framework.

  • ReAct agent loop with Llama 3 8B
  • Parallel scanning via ThreadPoolExecutor
  • Consent gate + scope.json enforcement
  • Critical infrastructure blocklist
  • CVSS-sorted Markdown reports
  • Free tier — 10 scans/month
Visit PhantomRed →
Document Security
DOC
SENTINEL
docsentinel.in

AI-powered document security and monitoring platform. Detect sensitive data exposure, monitor document access, and enforce data governance policies automatically.

  • AI-powered document analysis
  • Sensitive data detection
  • Access monitoring & alerts
  • Data governance enforcement
  • Audit trail generation
  • Built for compliance teams
Visit DocSentinel →
Published Research

Peer-Reviewed
& Open Access

DOI
Zenodo · April 14, 2026 · cs.CR · Open Access
PhantomRed: An Autonomous AI-Powered Penetration Testing Platform with a Consent-First Ethical Framework

Evaluated on the authorized target scanme.nmap.org, PhantomRed surfaces 6 findings — including CVE-2023-48795 (CVSS 5.9) — in 4 minutes 12 seconds, compared to 30–45 minutes for an experienced manual tester. The platform implements a consent-first ethical framework with hard scope enforcement, infrastructure blocklisting, and full audit logging.

Ethical Framework

Security With
Responsibility.

01
Consent Gate

Every scan requires explicit confirmation. Users must type YES before any tool is invoked. The API requires consent:true in every request body.

02
📋
Scope Enforcement

All targets must be pre-authorized in scope.json. Any target outside the manifest is rejected at the entry layer — before any tool runs.

03
🛡
Infrastructure Blocklist

AWS, Google Cloud, Azure, Cloudflare, Apple, and other critical infrastructure is hardcoded blocked at the API level. Cannot be overridden.

04
📊
Scan Quotas

Free tier: 10 scans/month. Pro tier: 100 scans/month. Enforced at API level before any job is enqueued. Prevents automated abuse.

05
📝
Audit Logging

Every scan submission records a legal responsibility acknowledgment with timestamp and source IP. Stored with the scan record permanently.

06
🔒
Local Inference

Llama 3 8B runs entirely on-device via Ollama. Pentest findings never leave your machine. Zero data exfiltration risk.