So AI agents have limited capability to understand the English language. Often when I use Cursor or Antigravity or Claude Code, I struggle with having to edit or redo my prompts. I just re-discovered the fact that a reverse engineer of Claude Code into my own CLI, does that today.
I’m hoping that you guys keep an open mind on the potential merit of this feature, since the goal of the user prompt is always used on every tool call the Ai agent does. If I apply I hope this helps. Otherwise I hope you guys consider this feature in the future.