New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Never Let Your AI Agent Touch PowerShell!
My desktop is very clean now; there's nothing on it.
I tried to recover the accidentally deleted files using Recuva, but unfortunately, I only managed to recover a small fraction of them. Who would have thought that this disaster stemmed from nothing more than a syntax error?
I end up burning the most tokens on the tug-of-war between AI agents and PowerShell. Honestly, WSL or a proper Ubuntu/Debian system is far better suited for coding tasks—even though installing the latest NVIDIA drivers on the latter is an absolute nightmare. 😇
Thanked by 1WyvernCo
Comments
They are powerful tools. It's tempting to let them run completely loose (Claude Code auto mode goes brrrr), because of the convenience. However, it might be wise to run them in a virtual machine. They can do a lot of damage regardless of OS you run.
That makes a lot of sense! I’m also considering whether it’s necessary to use Docker containers to constrain the AI agent.
Cool story, bro.
Skill issue.
which model was that?
I hope your PC catches on fire.
@Not_Oles
I think you use powershell quite a bit right?
good question.
Have you tried to get an answer?
I run AI agents on a vps. Would recommend.
Forget about AI agents, I shouldn’t be allowed to touch PowerShell!
I have a solution, but you aren't going to like it.
Do it yourself and avoid using AI for anything development related
Have you heard of the Recycling bin?
AI hallucinate a lot, so never install it on your main PC especially if you are new to this kinda stuff. Just like @concept said, use it on a VPS.
Agreed, it's too risky unless you made backups. If required, inspect the command first
I don't think it's wrong to install it on your Main OS and run it. The rule of thumb here is that YOU are responsible for what it does.
If it makes an error or hallucinates then it's up to you to add in guard rails.
If you make it run something that you are not inherently comfortable with an AI agent running autonomously, then you should NOT use it for that case.