See lots of comparisons to Ansible but Chef/puppet (both of which have agent-less modes) in Python instead of Ruby is what immediately came to mind. I guess Salt as well technically.
> I think it can sort of make sense for some people who sort of listen to music as a background noise
This is a whole genre of music in of itself that real people created which AIs were trained on stolen copies of to produce slop that doesn’t have to compensate the origins.
So it only makes sense in that case if you think slop derived from real art is awesome and that actual human beings can get fucked.
> For me personally, I don't really know. You can't just do the same thing because the economy is constantly evolving, but I can't see where it's going.
Considering it happens across both opencode and other apps like Claude and Codex as well as across models it seems like something inherent to the models themselves and not necessarily a bug in the apps wrapping them. But maybe there’s more opencode et. al could be doing to prevent it.
You’re not alone, I’ve absolutely seen the same behavior occasionally with Opus in OpenCode where it takes actions it shouldn’t be able to in plan mode.
“Plan” vs “execute” modes seem more like suggestions the models _mostly_ follow. I have absolutely had models (Codex and Sonnet/Opus) perform actions in plan mode they should never have been able to take like editing files or starting to work on a plan that was just created.
LLMs write like that because people write like that. That they used the rhetorical device three times in a row in a single paragraph makes me think it’s less likely to be LLM and just how that person writes.
“Borderline unusable” is such a hyperbolic way to describe a fully functional design that doesn’t happen to be responsive. Hacker News must be borderline unusable for you as well then, no?