CyberSentinel Phishing Detection Case Study
A collaborative full-stack application for analysing potentially malicious URLs
Overview
CyberSentinel is a collaborative cybersecurity project for phishing URL detection. A React frontend connects to a Flask backend that combines Random Forest classification, URL and content-based feature extraction, and VirusTotal checks. A companion Chrome extension presents security warnings while browsing.
The Problem
The project explores how multiple signals can help assess a suspicious URL. Instead of treating a URL’s appearance as proof of safety, the application combines model predictions with additional analysis and displays a result for the user to review.
My Role and Team
I contributed to the phishing-detection system and its companion Chrome extension. This was a team project, not a solo build. The repository credits Mohataseem Khan, Rehan Khan, Saad Shaikh and Ansari Husain as authors. The features described below refer to the shared application.
Technical Approach
A URL submitted through the frontend or extension is sent to the backend for analysis. The backend combines machine-learning output, VirusTotal checks and content analysis into a response for the interface. The repository also documents feedback review, model retraining, JWT authentication and QR-code URL analysis.
Implemented Features
Evaluation and Limitations
No numerical accuracy or production-security guarantee is claimed in this case study. An evaluation needs a documented dataset, test split and false-positive and false-negative analysis. Model predictions and third-party checks can miss threats or flag safe URLs; they should not be treated as proof that a link is safe.
Publication Record
The related research paper describes a broader security framework. This case study describes the URL-detection application in the public repository; the two should not be read as an identical implementation specification. Read the Cyber-Sentinel publication record at IJIRT.
