.png&w=256&q=75)
🤓 Latest Submissions
.png&w=828&q=75)
RepoPilot
Problem & Solution Statement: Unboarding onto large, complex, or unfamiliar codebases is a slow and error-prone process for developers. Understanding component dependencies, tracing execution call graphs, and predicting the ripple effect (blast radius) of changing a specific function often takes days of manual code tracing and debugger sessions. RepoPilot solves this by providing an automated, AI-driven developer onboarding and architecture exploration assistant: 1. Dynamic Repository Ingestion: Users can input any public GitHub repository URL or local directory path. RepoPilot performs AST (Abstract Syntax Tree) parsing to extract modules, classes, methods, functions, and code chunks into a structured symbol index. 2. Interactive React Flow Canvas: Instead of static diagrams, RepoPilot renders an interactive call graph and dependency visualizer. Developers can inspect symbol relationships, zoom/pan across modules, and filter by symbol type (Classes, Methods, Functions, Tests). 3. Blast Radius Impact Analysis: When a developer selects a symbol (e.g., modifying a helper function), RepoPilot traverses incoming caller chains using NetworkX graph analytics. It highlights direct callers, transitive upstream dependencies, and affected test files in real-time on the graph without losing total system context. 4. Repository RAG Assistant: An integrated AI chat assistant allows developers to ask architectural and functional questions (e.g., "How does prediction pipeline work?"). Using indexed code chunks and call graph context, the bot provides accurate, evidence-backed answers with precise file and line citations. Target Users & Uniqueness: Built for software engineers, tech leads, open-source contributors, and dev teams wanting instant architectural clarity. Unlike static documentation tools, RepoPilot combines graph theory, static AST analysis, and LLM-driven RAG into a unified, human-centered UI.
27 Sep 2026
