EverMind-AI/EverOS — reverse-engineered prompt

Build me a Python project for long term memory in AI agents. I want it to feel like a memory layer that any agent or chat assistant can plug into, so it can remember useful facts, user preferences, past conversations, and lessons learned across sessions instead of starting over every time.

Please include a simple local demo where I can chat with an agent, save memories, search them later by meaning, and show how the agent uses those memories in a later answer. Also include a few example use cases, like a coding assistant that remembers project context and a personal assistant that remembers ongoing goals.

I also want a basic way to compare memory approaches, with small benchmarks or evaluation scripts that show whether memories are useful, relevant, and not just random notes. Keep the setup clear, add a friendly README, and make it easy to run locally. Look up current docs online if you need to.