Smart Testing & Security Hub

Problem: In AI-assisted healthcare applications, such as clinical Retrieval-Augmented Generation (RAG) pipelines, system reliability and security are non-negotiable. However, writing comprehensive test coverage—specifically edge-case unit tests and adversarial security tests like prompt injections—is incredibly time-consuming. Developers and QA engineers often spend days anticipating edge cases for rate-limiting, vector database retrieval, and LLM manipulation. This manual bottleneck slows down deployment cycles and often leaves critical vulnerabilities undetected in production environments. Solution: To address this workflow challenge, I developed a "Smart Testing & Security Hub" workflow for the AI Clinical Decision Support Lite platform. Instead of writing tests manually, I utilized IBM Bob 2.0 as an autonomous QA engineer to deeply analyze the existing Flask API and RAG pipeline architecture. The solution automatically generated a robust, production-ready testing suite comprising 124 passing tests covering in-memory and Redis rate-limiting, ChromaDB retrieval edge cases, and complex adversarial prompt injection attacks. Target Users & Interaction: The target users are AI software engineers, QA automation engineers, and DevOps teams. They interact with this workflow directly within their IDE by invoking IBM Bob 2.0's agent mode. The developer provides a single, high-level prompt, and the AI autonomously writes, structures, and saves the pytest files directly into the project directory. Uniqueness & Effectiveness: What makes this creative and unique is the shift from "AI as a code autocomplete tool" to "AI as an autonomous testing architect." By utilizing parallel subagents, the workflow addresses both infrastructure stability and AI-specific security simultaneously. This automated workflow reduces testing time from days to minutes, eliminates manual oversight, and ensures high-stakes AI applications remain secure and resilient before deployment.