PhishGuard — ML Phishing Detector szerző: Muhammad Asghar

PhishGuard protects you from phishing attacks, malicious URLs, and deceptive websites using a machine learning model that runs entirely inside your browser — no server, no cloud, no data collection.
How it works
Every time you navigate to a page, PhishGuard extracts 30+ signals from the URL and runs them through a logistic regression classifier trained on real phishing datasets (PhishTank, OpenPhish). A risk score from 0–100% is calculated in milliseconds and shown on the toolbar badge.
Features
• Real-time scanning — every page navigation is analyzed automatically
• Color-coded badge — green (safe) → lime → amber → red → purple (phishing)
• Warning banner — a dismissible alert is injected at the top of high-risk pages
• Link hover tooltips — mouse over any hyperlink to see its risk score instantly
• Detection signals — explains exactly why a URL was flagged (typosquatting, IP address, suspicious TLD, fake HTTPS, etc.)
• Feature details panel — 12-signal breakdown per page
• Scan history — session log of every URL checked
• Manual URL checker — paste any link to analyze it on demand
What it detects
IP addresses used as hostnames, brand impersonation (typosquatting), deceptive @ symbols, URL shorteners hiding true destinations, executable file downloads, hex-encoded characters, suspicious TLDs, free hosting domains, data/javascript URIs, excessive subdomains, missing HTTPS, and more.
Privacy
PhishGuard collects absolutely no user data. All analysis happens locally using JavaScript. No URLs, browsing history, or any other information ever leaves your device.