Yesterday, when Agent v2 was released, I was so excited! For the past couple of months, it felt like a new spark—coding had become much smoother and more enjoyable. But when I woke up and got back to it, Agent v2 didn’t seem to be working. I provided the same instructions about six times, but it just got stuck and didn’t do anything. Has anyone else experienced this?
1 Like
Just opened a ticket about this, sorry! Looks like parsing is breaking for some thinking/analysis steps, should be hotfixed soon.
We’re shipping improvements continuously this week, that’s why agent v2 is early access while we work through kinks.
2 Likes
Brother Kody, let us know when the issue has been fixed. I tried again, and I’m experiencing the same issue.
1 Like
Can you send me a link to the repl where this happened so I can look at the trace?
1 Like
I tried again and the agent self acknolwedge the issue, he asked me to rollback and it seems working again back to his new self agentv2. Thanks!
1 Like
I was really excited, but after encountered so many miss-firing, I moved on. Disappointed!
Sorry about that, it’s still early access and literally shipping changes every couple hours, once it’s stable and generally launched hope you come back.
Is there a place to see all these shipped changes? I’d love to keep up and test things out.
There’s a changelog Matt Palmer emails out at the end of every week and is available here Replit Docs