CODE STAR

Every repository has a memory problem. Six months after the first commit, nobody can answer simple questions without a two-week archaeology dig: who owns this module, why is this bug still open, which dependency will collapse first, is the team burning out? Existing tools answer in fragments — a linter shows smells, an issue board shows tickets, a dashboard shows commits — but the fragments never meet, and nobody has time to stitch them together on a Friday afternoon. The result is slow onboarding, silent technical debt, surprise outages at 2 a.m. and developers who are exhausted before the sprint even ends. CODE STAR attacks exactly that: understanding a codebase should take one paste, not one sprint. Our solution is a developer intelligence platform that reads a repository URL once and returns five connected perspectives that finally talk to each other. The DNA Profiler fingerprints the codebase across Structural, Behavioral, Tech, Quality, Social and Temporal layers, giving scores, key metrics, side-by-side comparison against another repo and a weekly timeline of how the code evolved. Cold Case Files plays detective on stale bugs: every case gets a case number, severity, suspected cause, evidence count and a bug-density heatmap across files, plus a one-click AI-suggested fix written as concrete steps. Mood Ring makes the human side of engineering visible — a morale gauge, mood distribution bars, per-developer cards with commit load and quality, a 7-day trend and code-health KPIs, so leads catch burnout before it becomes attrition. Food Chain renders dependencies as a living ecosystem in React Flow, ranking apex predators, parasites and fragile links with risk scores and impact analysis when you click a node. 3D City Explorer rebuilds the whole picture as an interactive Three.js city where every module is a building: orbit it, zoom, search by name, click it and read LOC, language and complexity — plus satellite, street and traffic views. The interaction model is deliberately tiny: paste a repo URL, then generate one module or hit Generate All, which fires all five in parallel with Promise.allSettled and live progress bars. Every result lands in a shared report store, survives navigation, and exports as a print-to-PDF report — one input, five lenses, one artefact a manager can actually read. What makes it creative is the collision of worlds. Serious engineering data is staged inside a 1960s pop-art design system — sun yellow, signal red, process blue, ink-black borders, hard zero-blur shadows and ben-day halftone dots — while the city module turns dull metrics into a place you can literally walk through instead of a spreadsheet nobody opens twice. Efficiency comes from IBM watsonx.ai: every API route consults our IBM Bob client first and falls back to deterministic data, so answers stay fast, structured, explainable and demo-safe even when the network misbehaves on stage. Target users are engineering students, hackathon teams, startup leads and open-source maintainers — anyone who inherits code they did not write. They interact with CODE STAR in the browser: paste a URL, watch progress, explore, print. No CLI, no install, no configuration file to learn. The five modules also cross-reference each other, so a bug found in Cold Cases can be traced to its district in the 3D City or to a fragile link in the Food Chain, turning isolated reports into one continuous investigation. Within two minutes a newcomer knows what the project is, where it hurts, who is carrying it, what will break next and what it looks like from orbit — the exact knowledge that normally takes weeks of pair sessions, tribal memory and guesswork.