Hypabase is a Python library designed to handle hypergraphs, where a single edge (hyperedge) can connect any number of nodes, accurately reflecting real-world complexities that traditional graphs often oversimplify. It uniquely integrates provenance tracking, allowing every hyperedge to record its source and a confidence score, which is crucial for data originating from diverse systems like LLM extractions or clinical records. With automatic SQLite persistence, namespace isolation, and an O(1) vertex-set lookup, Hypabase provides a robust and scalable solution for structuring and querying intricate relational data, particularly in domains like knowledge graphs, AI agent memory, and RAG pipelines.
Key Features
01Hyperedges: Connects two or more nodes in a single relationship
0211 GitHub stars
03Provenance: Tracks source and confidence for every edge
04SQLite persistence: Automatically stores data in a local file
05Namespace isolation: Creates scoped views within a single database file
06MCP server: Integrates with AI agents via 14 tools and 2 resources
