Agent stuck "working" and when it works i'm in an endless loop.
I’ve been trying to build an app in Replit and I’m running into a brutal issue that’s making it basically unusable.
For about a week now, the AI agent has been constantly stuck in a “working” state indefinitely. Sometimes it’ll sit there for minutes or hours without producing anything. Other times it eventually outputs code, but then it starts repeating the same fixes, rewriting the same files, undoing previous progress, or cycling through the same errors over and over.
The frustrating part is that it works great for a little while, then suddenly becomes unusable mostly when I need something done quickly haha.
I've tried reseting the app, opening a new browser, using it on Crome, using it on my phone, private mode, but nothing helps.
I also cannot get into shell or connect via SSH the screen just loads. I was trying to tranfer it over to Claude two weeks ago but the GitHub tool wouldnt work - any insight on this would be super helpful as well.
Has anyone actually found a reliable workflow to prevent the endless “working” state and looping behavior?
Would appreciate any advice because right now I’m spending more time fighting the tool than building. I made a ticket but support has not responded to any of them so I guess they just go into a black hole.
Thank you!