Addressing review comments

Your PR is open. Someone reviews it — a teammate, an automated bot, maybe even Claude itself in a review pass — and leaves comments. “This could leak memory.” “Rename this for clarity.” “You’re missing a test here.” For a lot of people, this is the part that feels like a verdict. It isn’t.

Review comments are just another kind of prompt. Point Claude at them, it makes the changes, you push again, and the PR updates itself.

This is the exact iterate-with-Claude rhythm you already know from Lesson 5.3 — only now the feedback comes from a reviewer instead of your own eye. You don’t start over. You don’t open a new PR. You make the requested changes on the same branch, commit, and push — and the open PR automatically picks up the new commits. It just updates in place.

The Claude way

With gh set up (same tool as Lesson 8.6), Claude can read the comments off the PR directly:

address the review comments on this PR

It reads each one, makes the changes, and you push. Or, if you’d rather, paste a single comment in yourself and let Claude act on just that. Either way, a review comment becomes a prompt — and you already know how to work a prompt.

You still hold the judgment

Here’s the part not to skip: not every comment is right. A reviewer can be mistaken. Claude can misread what a comment was asking for. This is trust-but-verify and reading a plan critically all over again, just on the review side.

So a comment isn’t an order — it’s part of a discussion. You decide which ones to act on, and which to push back on in the PR conversation with your reasoning. “Good catch, fixed” and “I left this as-is because…” are both complete, professional answers.

What’s next

That’s every piece of the loop — commit, message, branch, push, PR, review. The last lesson runs all of it end to end on one real change, so the whole thing stops being seven ideas and becomes one motion. Next.