Could agentscope support durable memory across sessions?

Hi, I’m Vivek Gupta, Founder & CEO at [MemCode](https://memcode.in). AgentScope helps teams build understandable, trustworthy agents; MemCode could provide a transparent long-term memory component with explicit reads, writes, retention, and deletion.

MemCode could be integrated as an optional memory layer behind that experience: persist user preferences, decisions, summaries, and learned facts by user/workspace/agent; retrieve only the relevant memories at the next turn; and let operators control retention, deletion, and model/provider changes. The integration could start as a small adapter or API-backed memory provider, with no change to the existing model, orchestration, or self-hosted path. I’d be happy to prepare a PR around the extension point you prefer.

If the integration is useful to your users, we’d also be glad to offer a transparent 10% revenue share on MemCode usage attributable to the integration, subject to your project’s maintainer, disclosure, and governance rules; participation is optional and not a condition of the technical integration. Would you be open to a short call, or should I open a small implementation PR/discussion first? If GitHub Issues are not the right place for partnership proposals, I’m happy to move this to the channel you prefer.