Hacker News Reader

Comment by extr | Hacker News Reader

Why not? Have you ever actually used these things? The risk is incredibly low. I run claude code with zero permissions every day for hours. Never a problem.

View on HN

I have (not an exhaustive list) SSH keys and sensitive repositories hanging out on my filesystem. I don't trust _myself_ with that, let alone an LLM, unless I'm running ollama or similar local nonsense with no net connectivity.

I'm a few degrees removed from an air gapped environment so obviously YMMV. Frankly I find the idea of an LLM writing files or being allowed to access databases or similar cases directly distasteful; I have to review the output anyway and I'll decide what goes to the relevant disk locations / gets run.

They don't have arbitrary access over your file system. They ask permission for doing most everything. Even reading files, they can't do that outside of the current working directory without permission.

I'm pretty comfortable with the agent scaffolding just restricting directory access but I can see places it might not be enough...

If you were being really paranoid then I guess they could write a script in the local directory that then runs and accesses other parts of the filesystem.

I've not seen any evidence an agent would just do that randomly (though I suppose they are nondeterministic). In principle maybe a malicious or unlucky prompt found somewhere in the permitted directory could trigger it?

Comments like this just show how bad the average dev is at security. Ever heard of the principle of least privilege? It's crazy that anyone who has written at least one piece of software would think "nah, it's fine because the software is meant to ask before doing".

Your obviously skilled, spending the money on a Claude only machine would pay for itself in less than three weeks. If I was your employer, it would be a no brainer.

Make me that offer :D

You don't want to work in a billing system.

Reply on news.ycombinator.com