umbrel on Nostr: you might've noticed hermes and openclaw have just been… insanely reliable on ...

you might've noticed hermes and openclaw have just been… insanely reliable on umbrelOS lately, right?

here's why: both apps now ship with managed umbrelOS runtime context injected into the agent's prompts

the agent now knows it's running inside an umbrelOS app, where to correctly store data, that it shouldn't self-update, where the sandbox boundaries are, and when to load deeper umbrelOS-specific guidance

basically, the agent stops doing weird stuff because it finally knows where it is

you're welcome 🤝