With the rapid growth of Front-End technologies and the lack of code sanitization, multiple ways of attacking through code have emerged. Allowing cybercriminals to access confidential information, impersonate identities, steal money, access databases, among others. Our project approaches the 3 of 10 vulnerabilities identified by OWASP as the most recognized and usual in web applications, to identify and evaluate them at the JavaScript code level. For this, a machine learning model was proposed, detecting the frequency of occurrence of a word with the identified vulnerability; resulting in a model with an accuracy of 89%. Finally, an extension was implemented in Visual Studio Code to read in real time the code that the person is writing in order to identify which vulnerabilities it has.