Model-agnostic cognitive architecture for LLMs

Hi HN,A couple weeks ago I shared an early version of a side project I’ve been tinkering with called Persistent Mind Model. I built it at home on an i7-10700K / 32GB RAM / RTX 3080 because I was curious whether an AI could keep a stable “mind” over time, that could "think" about it's own identity as an LLM, instead of resetting every session.After a lot more tinkering, I think the architecture is finally in a solid place.Basically, it saves everything the AI does, thoughts, decisions, updates as a chain of events in a local SQLite database. Because the “identity” is stored in that ledger (and not inside the model), you can swap between OpenAI, Ollama, or other backends and it just keeps going from where it left off, and reasons about its own history/identity development.I cleaned up the runtime and added things like: a better control loop, a simple concept system for organizing ideas, graph-based telemetry so you can inspect how it evolves, and a draft whitepaper (for now), and several full sessions you can replay to see how the behavior develops.It's basically my experiment to develop persistent memories, and self-evolving "identities" for LLMs.The whole system is only a few MB of code, plus about 1.3 MB for the example ledger I’m sharing.If you’re interested in AI systems that can grow over time, or you want to experiment with persistent reasoning/memory/ verifiable mechanical cognition, I’d love feedback.Repo: https://github.com/scottonanski/persistent-mind-model-v1.0It’s open-source, free to use, and still early. but it’s already producing some surprisingly interesting results.Would love to see what others do with it.