Show HN: Rubi – A knowledge graph that cuts how much Claude Code re-reads

Category: ai-ml

Tags: knowledge-graph, context-management, ai-coding-agents, token-optimization, mcp

Score: 6.8/10 (Innovation: 6, Technical: 7, Documentation: 8, Utility: 6)

Rubi is a knowledge graph and context manager for AI coding agents like Claude Code, designed to reduce token costs by minimizing context re-reading and improving project navigation. It combines techniques like BM25, PageRank, and Personalized PageRank to build a searchable memory of code structure and decisions, with reported cost reductions from real usage studies.

Target audience: AI coding agent users, developers using Claude Code, backend devs, devops

Repository: https://github.com/YURIBEDRAN/rubi · JavaScript · MIT · 12 stars

View on Hacker News