PlugMem

READMEより

PlugMem PlugMem is a plug-and-play long-term memory system for LLM agents. Instead of storing and retrieving raw interaction histories, PlugMem organizes experience into compact, reusable knowledge units, allowing agents to recall what matters to agent decision-making with minimal context overhead. The module is task-agnostic by design and can be integrated into existing agent pipelines with minimal effort, serving as a general memory backbone for diverse environments such as dialogue agents, knowledge-intensive QA, and web automation. For more details, please see the full paper:…

GitHub上のプロジェクトREADMEからの抜粋です。

スキル

オープンソースプロジェクトです — コードを確認し、GitHubからセルフホストできます。