Mnemex
Provides AI assistants with a temporal memory system featuring human-like forgetting curves and local, human-readable data storage.
Mnemex is an innovative Model Context Protocol (MCP) server designed to equip AI assistants with dynamic, human-like memory capabilities. It implements a novel temporal decay algorithm, mimicking the Ebbinghaus forgetting curve, where memories naturally fade unless reinforced. Data is stored locally in human-readable formats—JSONL for short-term memory and Obsidian-compatible Markdown for long-term—ensuring privacy and transparency. The system features a two-layer memory architecture, smart prompting for seamless LLM integration, and a knowledge graph for robust entity and relation management, all while being fully configurable and Git-friendly.
