XNews

AsterMem: Open-Source Long-Term Memory System for AI Agents

AsterMem: Open-Source Long-Term Memory System for AI Agents

AsterMem is an open-source third-party memory system that provides long-term memory for any AI agent. - It accepts any format, including Markdown, text, and bookmarks. - Long content is automatically split into semantic chunks, tagged, and indexed with vector search. - Data can be read from Markdown files, databases, or vector stores, and backup is as simple as copying a folder. - It supports knowledge graphs, timelines, and vector spatial views for memory management. - Compatible with Cloud Code, Codex, Cursor, and other tools, and allows users to choose their own models.

Key insights

AsterMem Gives Users Full Ownership and Portability of Agent Memory

AsterMem provides a pluggable, open-source long-term memory system for AI agents. Unlike platform-bound memory solutions, AsterMem stores data in local folders, enabling simple backup and full user ownership. It supports semantic chunking, vector search, knowledge graphs, and timelines, and works with popular coding agents like Codex and Cursor. The model-agnostic design means users can select their preferred embedding and LLM models, avoiding vendor lock-in.

Source posts