Using an LLM to (re)write your prompt or system prompt (for local models) is free alpha.