Mnemex
Provides AI assistants with a human-like temporal memory system, featuring natural forgetting and reinforcement based on cognitive science principles.
소개
Mnemex is a cutting-edge memory system designed for AI assistants like Claude, enabling them to retain and recall information in a human-like manner. It addresses the common AI limitation of forgetting past interactions by implementing a dynamic memory model where information naturally decays unless reinforced through use, mimicking the Ebbinghaus forgetting curve. This intelligent system combines recency, frequency, and importance to determine memory strength, automatically promoting frequently used or critical information to long-term storage, all while ensuring user privacy by storing data locally in human-readable formats you control.
주요 기능
- Knowledge graph capabilities for entities and relations.
- Temporal memory with human-like forgetting curves and natural reinforcement.
- 4 GitHub stars
- Two-layer memory architecture (Short-Term and Long-Term) with automatic promotion.
- Local data storage in human-readable JSONL and Obsidian-compatible Markdown files.
- Smart prompting system for natural LLM integration, including auto-save and auto-recall.
사용 사례
- Empowering AI assistants to remember user preferences and past conversations across sessions.
- Providing long-term knowledge retention for AI-driven research and archival systems.
- Managing dynamic context for AI in rapidly changing or high-velocity interaction environments.
