I think this is too limiting. I hope you provide various context limits mode like:
  • All Messages: when the scenario is cumulative discussion with whole context needed
  • Without Context: when doing one-shot analysis with huge+complex system prompt
  • Last n Messages: like what currently available
  • First n Messages: when some of the first chats are important to carry on next input
Those scenario is needed in real life use case. This flexibilities are not exists in other apps, especially on web apps including ChatGPT. Your app will be the king of GPT client with this feature.