Hi everyone, I developed a project using caffeine. I have iterated 119 versions, and now I can’t deploy it. I encountered this error. I tried many times but still can’t deploy successfully. Is my development career over?
1.Generating backend code 
2.Creating a backend API for the frontend
3.Generating frontend code for your web app
Sorry, you hit an internal LLM error.
Unable to create your app
I’m very happy with using Caffeine, but there are some areas where it’s inconvenient. For example, I often need to locate the code location and tell Caffe to make the changes, but the current Caffeine website doesn’t allow me to quickly browse the code structure. Another problem is the lack of Git, so I don’t know exactly what Caffe has modified. Sometimes Caffe pretends to make changes, but in reality, they haven’t.
Caffeine has really given me a taste of natural language programming. Natural language combined with debugging capabilities can quickly adapt to this new paradigm, but we need data that can pinpoint the problem, and Caffeine needs to understand and address it precisely. I don’t mind slowly coaching Caffe, letting it help me with small changes, but it’s currently very inconvenient. For example, if a deployment fails, I can’t find the cause of the problem, and I can’t manually intervene to make changes.
”mylive-at8.caffeine.xyz“This is the project I generated after 119 iterations, but now it has stagnated. I still have a lot to achieve, but I don’t know where the problem is. It’s like a black box, and there is not even an error. There is only a cold “ Sorry, you hit an internal LLM error.”
In addition, if Caffe wants to grow into a groundbreaking product that allows users to develop complex projects, it should solve the problems that make users uncomfortable. I can accept bugs and the occasional logical confusion of Caffe, but you need to build auxiliary tools so that I can use Caffe as I please.
Same hit a block with LLM errors on all of my read project’s im trying to produce and they aren’t necessarily related as far as scope. Perhaps, equal in complexity. Now I’m at a stand still cant make any changes, can’t remove something simple junk in the code, can’t add something small. Just stuck at but I’m nowhere near 119 iterations. I’m impressed you got that far without experiencing these problems.
I plan to make a compile a list of roadblocks i hit where caffeine got stuck, and the prompts/solutions i used to eventually overcome the issues. Hopefully it helps somebody.
Great, my project has stagnated. I urgently need your help. I will be very grateful.
Dfinity needs to add more tools to caffe to help us debug, but it is very inconvenient at present. I hope someone can help us.
The Atlas milestone being ticked off is what you’re waiting for, will allow Caffeine-produced code to be exported/shared to ICP Ninja so you can dive ino the code manually. Projected to be hit sometime this month, but development is development so it could hit a snag. Or it could happen tomorrow.
If you are having technical issues and need help troubleshooting the problem please send a support email to support@caffeine.ai Thank you
I have found out what the problem is. My spec description has exceeded tens of thousands of words. Now the LLM loading spec has been launched, resulting in a display error. I hope caffe can fix this problem.
https://mylive-at8.caffeine.xyz
The solution is written here.
I’ve encountered a new problem.
Caffeine currently uses static file hosting and doesn’t yet support automatic redirection of all unknown paths to , resulting in 404 errors for SPA deep links. To enable direct access to any page, the platform needs to support “SPA route redirection” or “fallback to index.html” functionality, allowing all paths to be handled by the front-end router. You can follow the platform for future updates or contact us to report any issues. For now, we recommend that users access deep pages via the homepage redirect, and we’ll implement direct link access once the platform supports it.