CodeWeavers
Team consisting of an Intuit IAM Engineering Lead and an Accenture AI/ML Engineer skilled in RAG, Azure OpenAI, Python, and secure system architecture.
Project Description
CodeLens is an agentic generative UI application that turns a GitHub repository URL into an interactive architecture exploration canvas. Instead of stopping at a text chat response, the agent analyzes repository files, resolves imports, detects module relationships, and updates a live dependency graph that users can search, filter by language, inspect module details, and explore by architectural depth.
The project uses the Generative UI Global Hackathon starter kit with CopilotKit, AG-UI transport, A2UI-style generated tool surfaces, and a Python LangChain/LangGraph agent. The frontend is built with Next.js, React, and TypeScript, while the backend uses a CopilotKit BFF and a Python repository-analysis agent. The agent calls structured frontend tools such as setRepositoryGraph, selectModule, highlightModules, and setCanvasStatus, allowing its output to directly reshape the UI in real time.
CodeLens is original because it converts agent reasoning into a functional visual workspace for understanding software systems. Judges can see the agent move beyond conversational answers into a working, interactive graph interface that reflects the analyzed codebase, highlights important modules, and helps developers visually understand code structure, dependencies, and architectural flow.