for now i'm using activegraph myself in my projects, and sharing the use cases – while slowly updating the runtime based on feedback or as necessary i think this general approach will help us build longer running agents, and if people want to use my runtime, great. if all it