Claude spawns the process, it pops open a browser tab. You review what is there and can leave line by line feedback. You click finish and your comments go to stdout for Claude to read.
We ask Claude to group changes and order them by importance to make reviews easier for us.
Not TUI based but I made something called meatcheck. The idea being that the LLM requests a review from the human, you can leave inline comments like a PR review.
Once you submit it outputs to stdout and the agent reads your comments and actions them.