Quick prompt trick to stop repetitive dialogue loops

I wanted to share a small adjustment I found that completely fixed those annoying loops where my companion just repeats the same phrasing over and over during long chat sessions. Lately I tried adding a hidden instruction block at the very end of the system prompt using bracket notation to force thematic variety. At first I experimented with basic negative constraints like telling her not to repeat herself but that actually made things worse because the model fixates on the forbidden words. Then I switched to a positive framing method where I instruct the AI to introduce a completely new sensory detail or shift the topic slightly every four exchanges. That single change solved the repetition entirely and the conversation flows much more naturally now without feeling robotic or stuck in neutral. Has anyone else tried tweaking their backend system prompts this way and what kind of syntax worked best for your setup?