Skip to content

Sutharshannn/Neural-Trace

Folders and files

NameName
Last commit message
Last commit date

Latest commit

Β 

History

3 Commits
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 

Repository files navigation

πŸ›‘οΈ Neural-Trace: AI-Powered Binary Forensic Auditor

Neural-Trace is a sophisticated security auditing agent built to bridge the gap between low-level x86 Assembly instructions and high-level security intelligence. By leveraging the advanced reasoning of Gemini 3, it automates the detection of critical vulnerabilities in legacy binaries.


πŸš€ Key Features

  • Deep-Think Logic Simulation: Utilizes Gemini 3's thinking_level="HIGH" to simulate CPU register states and memory stack behavior, exposing the model's internal "thought process".
  • Agentic Vulnerability Mapping: Detects buffer overflows, integer overflows, and logic bombs with high precision.
  • Verified Exploitation: Uses Gemini's built-in Python Code Execution tool to mathematically verify memory displacements and potential overflow offsets.
  • Remediation Engine: Automatically generates memory-safe Python rewrites of vulnerable logic to assist in modernizing legacy systems.

πŸ› οΈ Built With

  • Python & Streamlit: For the orchestration engine and interactive forensic dashboard.
  • Google Gemini 3 API (Flash): The core reasoning and code execution engine.
  • x86 Assembly: The target forensic environment.

πŸ“– How It Works

  1. Ingestion: Upload a raw .asm or .txt binary dump.
  2. Audit: The agent performs a forensic scan, identifying insecure memory management.
  3. Verification: Using Native Code Execution, the agent calculates the exact byte-offsets required for privilege escalation.
  4. Reporting: A full forensic report is generated, including the AI's step-by-step reasoning chain.

πŸš€ Quick Start

  1. Clone the repo: git clone https://github.com/Sutharshannn/Neural-Trace.git
  2. Install dependencies: pip install streamlit google-genai python-dotenv
  3. Add your GEMINI_API_KEY to a .env file.
  4. Run the app: streamlit run app.py

πŸŽ“ Academic Context

Developed as a capstone-level project during the Gemini 3 Hackathon (Feb 2026). Author: Sutharshan Suthakaran


Disclaimer: This tool is intended for ethical security research and educational purposes only.

About

Forensic AI Agent using Gemini 3 to audit Assembly code for vulnerabilities.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors