Himan-D/agent-memory — reverse-engineered prompt

Build me a working version of Hystersis, an AI agent memory system that lets assistants remember conversations, facts, preferences, and relationships over time instead of starting from scratch every chat.

I want a Go backend with an API where users can create sessions, save memories, add chat messages, search by meaning, and give positive or negative feedback so the memory quality improves. It should support different memory types like conversation memories, semantic facts, knowledge graph relationships, and reusable skills. Please include local setup with Docker services for the databases, a simple CLI or examples for testing it, and an MCP server config so it can connect to Claude Desktop or Cursor.

Also include a basic dashboard or demo where I can compare an agent with memory versus one without memory, and see stored memories and search results. Keep the install and quick start simple, with sensible defaults and env examples. Look up current docs online if you need to.