Users Code Canvas saves chatgpt memory

Just an idea when chatgpt makes a canvas to place code on and the approx token limit is around 24k. This means GPT looses context of code. Soon forgeting function names etc. But if the code was on the users canvas and on their PC then gpt can read this back line by line or at users line.
At present gpt keeps a version of canvas itself and writes it back at an odd position on the users canvas. Making it become random or overwriting with new function names.

I can see that the developers tried to append code, functions etc. However append soon fails better that appended code starts with a remark/end remark so the users canvas just grows and grows. Which can be edited after the gpt can then use the context from the user at no cost to OpenAI in memory.