Anti-phishing techniques such as list-based filtering, website address and content feature analysis are continually improving along with machine learning models, computer vision technology, and natural language processing techniques. Efficient phishing detection often relies on a combination of these techniques. In this paper, we present a lightweight heuristic method for detecting phishing websites using URL and HTML features, and a prototype implementation as a Chrome extension. The proposed method consists of two parts: analyzing the HTML source code to determine if a confidential form exists; and designing a similarity algorithm to measure the correlation between the domain name and web page text information. The proposed solution eliminates the need for offline model training or third-party services, making it practical for real-time phishing detection. The experimental results demonstrate feasibility of integrating the proposed solution into a browser extension for efficient phishing detection.