Phishing is one of the most prevalent cybersecurity threats globally, targeting unsuspecting users by impersonating legitimate websites to steal sensitive data. This paper proposes a real-time phishing website detection system implemented as a browser extension, using a Random Forest classifier. The system extracts address bar, domain-based, and HTML/JavaScript fea- tures to classify websites as phishing or benign. Experimental results on a dataset of 6,000 websites yielded an accuracy of 86.1%, with a precision of 81.65%, recall of 93.05%, and specificity of 79.1%. The model is integrated into a Google Chrome extension, providing seamless and immediate phishing alerts to users during browsing.