Copilot code review drops its size limit and starts reviewing what bots wrote

GitHub changed three things in Copilot code review on August 27. Two of them point at the same situation: the pull requests being reviewed are increasingly not written by people.

The size cap is gone

Until now a pull request over 300 files or 20.000 lines of code fell outside what Copilot code review would look at. That limit has been removed, with no replacement figure named.

The cap mattered most exactly where review is hardest to do by hand. A migration, a dependency bump across a monorepo or a generated-code refresh clears 300 files without being complicated, and those are the changes a human reviewer skims.

Bot pull requests get the full review

Pull requests authored by bots, including GitHub's own Copilot cloud agent, are now covered. Ones from the cloud agent previously received a limited review and now get a full agentic one.

This closes a gap that was becoming awkward: the agent that writes the change and the reviewer that checks it are both Copilot, and until now the second one was doing less work precisely when the first one had done all of it. Organizations that allow members without a Copilot license to use code review on GitHub.com can have bot-authored pull requests reviewed too, billed to the organization.

Resolving a review comment now takes a reason from a dropdown next to the resolve button: addressed, won't fix, or incorrect. That third option is the useful one, because it is the only signal in the flow that separates a suggestion a person disagreed with from one that was simply wrong.