Anyone actually building persistent agent behavior?? existential crisis rant, tnx.

About this article

Been grinding on this solo since aug, a behavioral spec layer for AI agents — personality persistence, state machines, emotion systems preetty much ai thats not shitty. JSON spec that the model interprets directly. " But don't worry that's just a prompt or theatrics" LLM getting better at agentic tasks?? Weird right..... ACP, A2A, MCP — those are transport. This is what the agent actually is Definitely needs testing though there's a potential it might be actively to a degree shifting how the ...