The development and maintenance of RESTful APIs often suffer due to time-consuming and error-prone manual documentation processes, which ultimately affect testing quality. RESTSpecIT is an innovative approach utilizing Large Language Models (LLMs) to infer RESTful API specifications and conduct black-box testing efficiently. The technique employs an in-context prompt masking strategy that requires no model fine-tuning and demonstrates impressive capabilities in inferring GET routes, discovering undocumented routes, and identifying server errors.

• RESTSpecIT identifies 85.05% of GET routes and 81.05% of query parameters on average.
• It uncovers undocumented and valid routes and parameters, enhancing API robustness.
• The tool exposes server errors, contributing to overall API reliability.

LLMs’ knack for automating tasks is brilliantly tapped into with RESTSpecIT, easing the burden of documenting and testing RESTful APIs. Its potential for minimizing human error and boosting the efficacy of API management is significant. In future work, this model could be expanded to cover more complex API interactions and potentially integrate with continuous integration/continuous deployment (CI/CD) pipelines. Find out more about RESTSpecIT.