Problem with not providing full Code

bolt.diy [bolt.diy] Issues and Troubleshooting
1

The last days I had more often the problem that bolt not writing all code to the files or not getting it from the LLM. As far as I remember I didn´t have this problem a week ago and now very often. Maybe its also cause I´ve got a bigger code project?!

image
image749×411 12.8 KB

image
image1115×618 44.2 KB

2

which LLM you are using and how many files you have in your project ?

3

Happend with multiple LLMs. Tried switching them.

Mostly use these with OpenRouter:

  • Qwen2.5 Coder 32B
  • LLama 3.3
  • ChatGPT 4o (but this is already not working anymore and says to much okens)
  • Claude 3.5 Sonnet
  • Gimini Flash 1.5 (think the screenshot was from this one)

Projectfiles:

image
image313×647 13.2 KB

4

Just did a quick test again:

image
image1669×894 157 KB

5

and how long is your chat ??

6

Switched to OpenAI and got the error on token limit. So this should answer the question (54k)?

image
image1896×815 89.1 KB

7

I am guessing it has a too long chat history going on and all of them are going into the context causing the context to overload

8

Thought the same but this makes no sense when using Gemini with 1000k context, right? it should fit?