Ai Based Cyber Security Assistant For Non -Technical Users

Abstract

This project presents the development of an intelligent,
offline-capable cybersecurity platform called Cyber-
Wraith, integrated with AI-powered phishing
detection, password risk analysis, and simulated
breach monitoring. The system uses SQLite for local
storage of scan history and user preferences, while the
front end is built with React and packaged into a
desktop application via Electron, offering a
responsive, modular, and privacy-focused user
experience. Cyber-Wraith simulates real-time threat
detection by analyzing URLs, domain metadata, and
password strength using rule-based logic and entropy
scoring to identify security risks.
To enhance interactivity, the platform integrates a
local AI assistant (Ur-Luna) using the GPT4All model,
enabling users to receive natural language
explanations, guidance, and threat insights without
requiring internet connectivity. The system also
simulates breach alerts and domain risk scores using
predefined data and offers auditory alerts via the
Speech Synthesis API, improving accessibility. A
simulated logic layer calculates phishing
probabilities, password weakness, and domain trust
levels using heuristic indicators such as suspicious
keywords, SSL status, and metadata age.
The React-based dashboard features scan history,
animated risk indicators, and modular scan tools for
phishing, passwords, domains, and chat, with
preferences for dark mode, alert control, and voice
interaction. Real-time charts and risk summaries
display user insights in a visual format. This intelligent
platform strengthens cybersecurity awareness through
AI-driven, offline automation—empowering users
with local tools to identify and understand digital