Claude-Mem (github.com)

Claude-Mem, a new persistent memory compression system for Claude Code, has been announced, enabling the seamless preservation of context across coding sessions. By automatically capturing tool usage observations and generating semantic summaries, Claude-Mem allows developers to maintain continuity in their projects, significantly improving productivity. Users can initiate a new session with simple terminal commands, and previous context will automatically be available, reducing the token costs involved in retrieving relevant project history. This innovation is particularly significant for the AI/ML community because it addresses the challenges of context limitations often faced in traditional coding sessions. Claude-Mem's features include a smart search capability that utilizes a natural language interface, allowing users to quickly query project history with substantial token savings—up to 2,250 tokens per session. Additionally, its beta feature, Endless Mode, enhances performance by compressing outputs and enabling a linear scaling of tool usage, thereby preventing context exhaustion. With privacy controls, context configuration, and an intuitive web viewer interface, Claude-Mem promises to be a transformative tool for software developers leveraging AI in their workflows.