Shodh Memory

Shodh Memory is an open-source, neuroscience-inspired, offline memory system for AI agents. Unlike typical vector stores or LLM-based memory managers, Shodh Memory uses Hebbian learning and decay curves, strengthening connections as memories are revisited and allowing unused information to fade naturally, resulting in organic, human-like memory for AI. It is designed for speed (microseconds per operation), privacy (runs entirely locally with no cloud), and integrates natively with Model Context Protocol (MCP) for agent frameworks. Ideal for developers building autonomous AI agents, robotics/IoT solutions, or privacy-focused applications.