neo4j-labs/agent-memory ? reverse-engineered prompt

Build me a Neo4j backed memory system for AI agents that can remember conversations, store useful facts about people and projects, and keep track of the reasoning behind decisions and tool use.

I want it to work both as a Python SDK and a TypeScript SDK, with the same memory model in each so agents in either language can share the same stored memory. It should support short term conversation history, long term knowledge graphs, entity extraction and deduping, and ways to retrieve past reasoning or similar tasks later. If it makes sense, include a simple hosted style API flow as well as a direct Neo4j connection flow.

Please also add a few runnable examples that show how an agent saves a chat, updates memory, and then gets that context back in a later session. Use current docs online if you need to, and make sure the code is clean, testable, and easy for someone to copy into their own app.