Testing Prompt Variants in Conversations and Agent Configurations

Hi all. Is there a way to test variants of prompts in a conversation, similar to whats possible e.g. in Claude or Anthropic Console? This would be great in both a conversation (there does not seem to be a way to edit historic prompts at all?) and when building an agent (almost like a Sandbox before release). I understand the process right now would be to adapt the settings, prompts etc for the custom agent, then play around, test the output, go back into settings, change, save, go and test again (loop this process). Any other suggestions of how to test prompts in conversation and agents config?