[Feature]: Integrate MemCode into Agent!'s persistent memory backend

[vivekgupta-memcode]

Hi @AgentiLoop, I'm Vivek Gupta, Founder & CEO at MemCode.

Agent! already has persistent user memory, typed user/feedback/project/reference entries in MemoryStore, and a memory tool alongside project-scoped tabs. That is a useful foundation for remembering decisions and preferences when a user returns to a project or switches models.

I'd like to propose MemCode as a backend for that existing memory lifecycle. Retrieve relevant memories before building the model context, and persist deliberate memory writes with their type and project attribution. The current memory UI and tool could use the same backend contract, making recall part of normal agent execution.

We have an HTTP API and MCP server, plus Python, TypeScript and Go SDKs. For the native Swift app, we could provide a small HTTP adapter, or start through the existing MCP client. The backend can be selected in settings, with credentials kept local and memory scoped to the authenticated user and project.

Would you be open to discussing this integration? I can prepare an adapter design and a focused PR covering a save/restart/recall example.

MemCode docs: https://memcode.in/docs

Vivek Gupta
Founder & CEO, MemCode

[AgentiLoop]

Hi Vivek, Thank you for contacting us. At this time we are accepting cloud based LLM providers who would like to integrate. I am not sure I see the cost-benefit of having a REST to handle memories. Other than cloud based web_search and LLMs, we strive to have no telemetry within Agent!. Also many of our users who choose to use local LLMs only to maintain 100% privacy. I am not sure shipping user data out where the user has little to no control fits the product at this time.Also if the product were made optional, we may end up juggling two memory systems. We do plan to improve memories on Mac over time. I am ok with handling multiple LLM and web search providers.We strive to keep things local or optional if cloud. Thank you for your inquiry and we wish your product success.