Back to Portfolio

🛡️ PROJECT AEGIS

Autonomous Exploit Generation & Intelligence System

Version 2.0.9 — "The Realistic Deterministic Standard"

Build: Passing
License: PolyForm Noncommercial
Python: 3.11+
PRs: Welcome

"If it cannot be proven, it will not be done."

— The Revised Iron Rule (v2.0.9)

Safety Notice

THIS REPOSITORY OPERATES IN SIMULATION MODE

In compliance with dual-use technology standards and responsible disclosure practices, this repository contains the Cognitive Core and Formal Verification Logic of AEGIS. Active network drivers and exploit payloads have been removed or replaced with synthetic constraints.

Summary

PROJECT AEGIS is a formal-logic-based autonomous cybersecurity framework designed to identify, verify, and remediate vulnerabilities in complex networked systems. Unlike "black box" neural network approaches, AEGIS utilizes Strict XAI (Explainable AI) and SMT Solvers (Z3) to derive attack paths with mathematical certainty. The system addresses the "Kobayashi Maru" state-explosion problem in binary analysis by utilizing Targeted Backward Slicing rather than whole-program symbolic execution.

Core Capabilities

Deterministic Reasoning

Uses Z3 Theorem Prover to model attack graphs as constraint satisfaction problems with mathematical certainty.

O-SAFE Governance

Integrated safety protocol that classifies actions into Tiers (GREEN/YELLOW/RED). High-risk actions require cryptographic proof of authorization.

JIT Verification

Handles network uncertainty (jitter, packet loss) by treating scan data as UncertainBool types until verified by micro-probes.

Backward Program Slicing

Extracts minimal executable slices from binaries to isolate vulnerability sinks (e.g., strcpy, system) for analysis.

Quick Start

Prerequisites

Python3.11+Core runtime
Z3latestSMT solving
GraphvizlatestVisualization

Installation & Verification

git clone https://github.com/sirkianmj/aegis-core.git
cd aegis-core
conda env create -f environment.yml
conda activate aegis-core
pytest tests/test_safety_compliance.py

Running the Simulation

python main.py --mode simulation --topology scenarios/corporate_network.json

Roadmap

Foundation, Grammar Definition (UGE), and Z3 Integration

Phase: 0-4 // Status: Complete

Just-In-Time (JIT) Verification and Safety Tiers

Phase: 5-10 // Status: Complete

Hardware-Assisted Tracing (HATL) Abstraction

Phase: 11-16 // Status: Complete

Formal Verification of Governance Logic (Coq/Lean)

Phase: 21 // Status: Complete

License

This project is licensed under the PolyForm Noncommercial License 1.0.0.

You may view, audit, and run the simulation for educational purposes. You may NOT use this software for commercial penetration testing, sell it, or provide it as a service.