My first thought was, "meh, I already have Claude Code". But then I remembered my primary frustration with Claude Code. I need other LLMs to be able to validate Claude Code's assumptions and work. I need to do this in an automated way. Before Cursor CLI, I did not have a way to programmatically ask Cursor do this. It was very manual, very painful. But, now I can create a Claude Code agent that is a "cursor-specialist" that uses cursor cli to do all of that in an automated way.
Interesting, are you saying you would setup a Stop Hook in Claude Code that calls the Cursor CLI to have it validate and prompt Claude Code with further instructions?
