YEETUZ‑X444
· GODMODE

An autonomous pentest suite with a GUI — recon, exploitation, and reporting orchestrated from one control surface. Multi-model LLM core, real tool wrappers, full history. Built for authorized engagements, end to end.

COMING SOONprivate build · not yet released

The engagement pipeline

One control surface, four phases

Web GUI on Flask + Socket.IO over a Python backend. Every phase logged to a searchable SQLite/FTS5 history and rolled into a report.

PHASE 01

Recon

nmap, gobuster, nikto — orchestrated, not copy-pasted

PHASE 02

Analyze

DeepDiveAgent reasons over web-app surface

PHASE 03

Exploit

sqlmap + payload library, behind confirmation gates

PHASE 04

Report

findings → clean, reproducible output

What's under the hood

Multi-model LLM core

Claude, OpenAI-compatible, or local Ollama behind one call_llm() — auto-selects the strongest configured provider.

llm_providers.py

Real tool wrappers

nmap, sqlmap, nikto, gobuster/gotor wired to a control surface, plus a custom payload library.

wrappers + payloads

Searchable history

Every action stored in SQLite with FTS5 full-text search, feeding the report generator.

db.py · FTS5

Autonomous engine

Multi-agent runner with a DeepDiveAgent — heading toward a reasoning loop with verification-by-exploitation.

auto_pentest.py

Authorized use only

GODMODE is built for legal, scoped engagements and bug-bounty work. The guardrails aren't decoration — they're wired into the tool.

Scope-guard and target allowlist — it won't touch what you didn't authorize
Confirmation modals before any destructive action
Full audit trail — every step is logged and attributable