Web-based applications has turn out to be very prevalent due to the ubiquity of web browsers to deliver service oriented application on-demand to diverse client over the Internet and cross site scripting (XSS) attack is a foremost security risk that has continuously ravage the web applications over the years. This paper critically examines the concept of XSS and some recent approaches for detecting and preventing XSS attacks in terms of architectural framework, algorithm used, solution location, and so on. The techniques were analysed and results showed that most of the available recognition and avoidance solutions to XSS attacks are more on the client end than the server end because of the peculiar nature of web application vulnerability and they also lack support for self-learning ability in order to detect new XSS attacks. Few researchers as cited in this paper inculcated the self-learning ability to detect and prevent XSS attacks in their design architecture using artificial neural networks and soft computing approach; a lot of improvement is still needed to effectively and efficiently handle the web application security menace as recommended.