Hello everyone,
I am using Cursor for AI editing and conversations, and I am interested in exploring existing ready-made solutions for cross-session “infinite memory”, where conversation context or memories are stored as vectors so that new sessions can semantically retrieve only the relevant memories without transmitting the entire full history.
Since I prefer local deployment, my environment is an Apple M1 laptop with average performance. I’m looking for a free or open-source MCP Memory Server solution that can run locally and deliver good performance. My main requirements are:
- ARM64 Native Support: Able to run directly on M1 or provide a compatible Docker image.
- Low Maintenance: No complex cluster management; easy to set up and maintain.
- Local Persistence: Support local storage methods such as SQLite or file system, with no need for additional cloud storage.
- Additionally, I would like to know: why hasn’t Cursor officially implemented this cross-session memory feature yet? Is there a roadmap or technical challenges?
Thank you very much for your insights and advice!