This article presents the development of an API in NodeJs to automatically extract parameters in EPUB, DOCX and PDF digital learning objects. These parameters are used to check the degree of accessibility of Digital Learning Objects for people with hearing and visual impairments. The API uses refactoring techniques and the use of recursive functions, resulting in a JSON object that can be used by many programming languages. The article presents the results of preliminary tests, where it was found that EPUB files are better structured, leaving the parameters completely free to be extracted. The validation will be supported by the Development and Innovation group in Rehabilitation and Digital Assistive Technology (DIRTA-UFSM) and the accessibility team at the Social Media and Social Inclusion (LAMID) laboratory at the Federal University of Santa Catarina (UFSC).