🤓 Latest Submissions
Code Noir
Developers often lose valuable time when joining an unfamiliar project because they must manually inspect folders, configuration files, dependencies, entry points, architecture, and undocumented conventions before they can contribute confidently. In hackathons, internships, open-source work, team handovers, and fast-moving development environments, this onboarding cost becomes even more significant because documentation is often incomplete, outdated, or scattered across the repository. Code Noir solves this by turning an unfamiliar codebase into a structured developer onboarding experience powered by IBM Bob. The user uploads a project as a ZIP file. The Spring Boot backend safely extracts the repository and invokes IBM Bob through Bob Shell. Bob inspects the uploaded codebase and identifies the project purpose, primary language, frameworks, build tools, architecture, important folders, key files, noteworthy findings, recommended reading order, and a practical starter task for a new developer. Instead of displaying a long, unstructured AI response, Code Noir normalizes Bob’s analysis into a consistent language-agnostic JSON report. The report is parsed by the Java backend and rendered through a polished Thymeleaf interface using compact cards, expandable details, severity indicators, and focused onboarding sections. This keeps the experience readable while still preserving detailed technical information when the user wants to explore further. Code Noir is designed for developers who need to understand a repository quickly: new team members, interns, students, hackathon participants, maintainers taking over legacy projects, and contributors exploring unfamiliar open-source code. The interaction is deliberately simple: upload the repository, wait while Bob analyzes it, and receive a guided map of where to start. A key design goal is language independence. Code Noir does not hardcode separate analysis logic for Java, Python, or other stacks. IBM Bob performs the repository understanding, while Code Noir enforces a common report structure. Core fields remain consistent across projects, while stack-specific information is stored as flexible technical metadata. The prototype was tested with both a Java Spring Boot project and a Python Flask project, demonstrating that the same onboarding workflow can adapt to different technology stacks. What makes Code Noir unique is the role of IBM Bob. Bob is not used only as a coding assistant during development; it is part of the application’s runtime intelligence layer. The application orchestrates Bob’s repository analysis, structures the output, and presents it as a developer-focused onboarding experience. By transforming repository exploration into a guided workflow, Code Noir helps developers move from “Where do I start?” to an informed first contribution faster, with less manual searching and less dependence on incomplete documentation.
27 Sep 2026
