CodeRescue AI :Legacy Code into Resilient Systems

### Problem & Solution Statement Modernizing a legacy application is not simply a matter of generating new code. The real challenge is understanding what already exists before changing it. Large legacy codebases often contain tightly coupled modules, undocumented dependencies, inconsistent error handling, security risks, duplicated logic and critical functionality with little or no test coverage. Developers can spend significant time trying to understand the system, while an apparently small change can unexpectedly break another component. A complete rewrite is often expensive, risky and impractical. **CodeRescue AI** is designed to solve this problem by transforming legacy-code modernization from a guessing exercise into an evidence-driven rescue process. Our solution uses a real e-commerce application, **ShopSphere**, as the system under analysis. Instead of relying on manually entered or fictional findings, CodeRescue AI examines the actual project structure and source code to discover files, modules, routes, functions, dependencies and application relationships. This information is transformed into a **Rescue Map**, giving developers a visual and structured understanding of how the application works. The platform then identifies meaningful **risks, dependency relationships, critical components and test gaps**. Findings are connected to the underlying code wherever possible, allowing developers to understand not only *what* is wrong, but also *where* it occurs and *why* it matters. The next stage is the **Bob Rescue Workflow**, where IBM Bob acts as an AI-assisted engineering partner. The workflow follows: **Investigate → Analyze → Plan → Modify → Test → Review → Report** Rather than blindly replacing legacy code, the system encourages targeted, traceable improvements. Developers can investigate an issue, understand its impact, create a remediation plan, implement the change, validate it through testing and review the resulting evidence. The target users are developers, students, engineering teams and organizations responsible for maintaining or modernizing existing software. What makes CodeRescue AI distinctive is its **“rescue, not rewrite” philosophy**. It connects code analysis, architecture understanding, risk discovery, dependency mapping, testing gaps and AI-assisted remediation into one continuous workflow. The final **Rescue Report** provides a before-and-after view of the modernization journey, making the process understandable and traceable. CodeRescue AI aims to make legacy software **understandable before it is changed, safer while it is changed, and measurable after it is changed.**