Hacker News new | ask | show | jobs
by chris_money202 39 days ago
I prefer GitHub Copilot Agent Session for this flow since it's in its own container/VM naturally and you can spawn from the Github Mobile app, except it is a bummer how much of your actions minutes that and CI/CD will consume.

If I were to do this Codex flow I would want to have it setup in a dev container, most people are probably not going to do that, so we are going to see a lot of vulnerabilities introduced / computers "breaking". Breaking in quotes because the computer is not actually broken, but to a novice it might appear it is, when in reality it's just out of disk space or the agent executed a setting, it shouldn't have. Unfortunately, if the computer is out of disk space, a novice won't be able to spawn their coding agent to fix it, so their next logical course of action is Geek Squad/IT? I don't even know.

1 comments

No we won't see that because the whole 'broke my computer' trope pretty much never happens to people unless they are researchers going wild in a VM to trigger it. And pure novices are just... not using these tools, and when they do, at most they lose a spreadsheet. (and that hardly happens either, if at all)

This is a non-issue problem raised by anti-ai zealots, much like data center water use is overblown. Headlines and lies.

I am an engineer and just the other day forgot to set a disk quota on a container I gave an agent and it almost reached disk space running a simulation that was logging to a DB overnight. If that would have happened I probably would have lost ability to ssh and that computer isn't connected to a display or keyboard so would have been a huge headache for me to reset.

And novices ARE absolutely using these things, I have a handful of friends with 0 CS background using claude to write apps, automation, etc

Also to drive this point home I've had claude suggest no password sudo before, some people are going to blindly accept that on a computer that also has access to accounts. Making that computer very vulnerable.
Neither of these (yours and other commenter) points drive anything home. Both involve an experienced dev dealing with systems they understand, avoiding "issues" due to being experienced. I say again, show me the actual accidental catastrophe caused by a novice using agentic coding tools. You take for granted the barrier to entry here. You take for granted what a colossal minority we are in, those of us who use these tools. What a minority it is to even have ever opened a CLI to install something. And even then, the self risk factor is so miniscule. There are fools (admittedly ill-advised even from my perspective) who have their claudes on ralph loops for days... and theyre all doing... just fine.
I am still waiting for the story where it doesnt involve an experienced dev who is perfectly aware of the risks, pushes the boundary, and makes a mistake....and then catches it before it gets out of hand. You're literally proving my exact point. Novices are not SSH'ing into servers. The very act of having something to SSH into eliminates potentially the entire novice pool.

Also, your worst case was... having to dig for some cables and peripherals, which of course you have around, because you are into computers and self hosting.