Extracting REST API Endpoints from Microservices Using LLM Agents

This paper uses LLM agents to explore automated API endpoint extraction and OpenAPI documentation generation from Node.js microservices. The proposed system analyzes microservice source code, processes it, and extracts REST API endpoints with its data. As a result, the system provides a list of endpoints for the microservice and generates an OpenAPI file that can be used for further analysis. The system design and prototype were implemented, tested, and evaluated. Results show that the automated system reduces the time needed to prepare microservice documentation and helps the developers maintain it without managing routine tasks like writing and managing the documentation file or the comments in the source code. The LLM agentic approach provides a foundation for further automated source code analysis and generation, especially when big LLM models are used.