Extracted Agent Memory from OpenAI Agents into a reusable and standalone library

1 point

8 months ago

I extracted the memory management functionality from OpenAI's Agents framework into a standalone Python library that anyone can use in their AI projects.

Key features: • SQLite & OpenAI-based conversation storage • Proper session management • Clean API with convenience functions for quick setup • Comprehensive examples and batch processing tools • Zero dependencies on the original Agents framework

The library handles conversation history, context memory, and provides export capabilities in JSON/text formats. Perfect for chatbots, AI assistants, or any application that needs persistent conversation memory.

Originally buried inside a larger framework, this memory system is now available as a focused, well-documented library. All tests pass, examples work out of the box.

GitHub: https://github.com/LLMNativeOS/smem