GitNexus: code knowledge graphs
Tech with Mak highlighted GitNexus, a tool that converts codebases into queryable knowledge graphs for impact analysis, execution flows and refactors — a strong portfolio idea for backend and infra candidates. The project demonstrates practical static analysis and graph modeling skills that interviewers at big tech care about. (x.com)
GitNexus debuted as a zero‑server, client‑side code intelligence engine with a live web UI at gitnexus.vercel.app that accepts GitHub URLs or ZIP uploads and runs processing entirely in the browser. (gitnexus.vercel.app)) The project’s pipeline indexes repositories into a property-style knowledge graph and has published example graphs with roughly 13,000+ nodes and 38,000+ edges representing functions, classes, imports, call chains and execution flows. (nathan-norman.vercel.app)) Storage and query infrastructure for indexed graphs in some builds use KuzuDB as the property graph backend, while parsing and local analysis run via a client-side pipeline that avoids sending source code to remote servers. (deepwiki.com)) GitNexus exposes the graph to tooling through a CLI and an MCP-compatible interface for editor/agent integrations, and the project README documents commands like gitnexus analyze and gitnexus setup for registering agent hooks. (npmjs.com)) Authors and early coverage emphasize the product goal of reducing “blind edits” by AI assistants by surfacing downstream blast radius and dependency chains, positioning the tool as a bridge between static analysis and graph‑RAG agent workflows. (deepwiki.com)) Public uptake spiked on release—multiple community writeups and aggregated trackers report rapid star growth on the project's GitHub and active discussion in blogs and tutorials that demonstrate local, privacy‑preserving analysis for proprietary codebases. (rywalker.com))
