BioRAG
A biomimetic, symbolic cognitive memory architecture for AI agents that gives them persistent cross-session memory using neurobiology-inspired structures and linear algebra instead of dense embeddings.
BioRAG addresses the amnesia large language models have between sessions. It represents memory with a hybrid PPMI-and-SVD factorization, a Hebbian synaptic graph, sparse distributed memory, conformal calibration and ACT-R-style activation, rather than dense vector embeddings, and its search path needs no GPU and makes no external API calls.
It is bilingual in Spanish and English, with stemming and symbolic expansion through WordNet and a set of concept hubs, and is organized into modular submodules for its memory core and for its MCP server.
It is written in pure Python with SQLite FTS5 storage.