I’m encountering consistent retries failed in prompt response. I tried to clean cache but it doesn’t seem to be a right fix.
-Is there a place I could troubleshoot to see if I’m under a limit or facing a connection issue?
In my own experience, this has happened when asking it to work a very complex problem that involves a lot of information and/or has complex logic. Try simplifying your prompt, or directing it to focus on only a part of the problem. Make sure your GEMINI.md and/or .idx/airules.mddon’t require it perform complex analysis.
1 Like
I did some random easy tests in prompt like “change version number by plus one” and it still failed. Currently it doesn’t feel that I’m hitting a quota limitation wall or maybe it’s an underlying rule to prevent my continuous work on it.
But anyway I’m thinking of switching to different agentic toolkit.
Thanks.
happens daily regardless of the prompt complexity. extremely annoying
This is very annoying and 100% understand your frustration.
Things I have done to jar it lose is start a new chat and type something and get it to stop there and once it starts to work again I switch back to the other chat.
Also keep your chats very small and do not try to make everything happen for your project with one continuous chat with the code assist.
The longer the chat session, the more prone to errors.
It can also simply be a lot of activity on the bot as well and it simply didn’t have the ability to fully run your ask.
but I 100% agree, it is annoying but work arounds can help and following the advice of airules in your .idx and the GEMINI.md in the root. please note that once you make these you have to start a new chat for the rules to take affect or tell your cli to read those files by providing the relative path. Frankly after like 4-5 edits, I remind the bot to read the rules just to keep it in check because the longer the chat gets, the more it seems to distance itself from the rules you have in place.