GitHub’s Copilot code review has been updated to cover two previously unsupported pull‑request scenarios and to expose a new comment‑resolution dropdown. The changes let the service automatically review PRs created by bots—including the Copilot cloud agent—and remove the former 300‑file/20,000‑line size cap, while also allowing reviewers to tag a resolved comment as Addressed, Won’t fix, or Incorrect.
Expanded PR coverage for bot‑generated changes
When a pull request originates from a bot, there is no individual Copilot‑licensed user to attribute the review to. Enabling the “Allow members without a Copilot license to use Copilot code review in GitHub.com” policy now permits the organization to bill the review usage directly, extending full‑agent reviews to bot‑authored PRs and to those opened by the Copilot cloud agent. This eliminates the fallback to a limited experience that existed before.
Removal of large‑PR limits
Previously Copilot code review stopped processing PRs that exceeded 300 files or 20,000 lines of code. The limit has been lifted, so the service will attempt a full review regardless of size. Teams can therefore rely on automated feedback for massive refactors or bulk updates without manually splitting the work.
Resolution reason dropdown
A new UI element appears next to the “Resolve conversation” button. Selecting one of the three reasons—Addressed, Won’t fix, or Incorrect—sends explicit feedback to the product team. This data can help improve the underlying model and gives engineers a lightweight way to document why a comment was dismissed.
Related CloudNinjas coverage: AI engineering.
What This Means For Practitioners
AI and cloud engineers should verify that the organization‑wide policy is enabled if they rely on bot‑generated code, as the change shifts billing to the org level. DevOps and SRE teams need to adjust monitoring to capture any increase in Copilot usage now that large PRs are no longer filtered out. Security engineers should treat the new resolution metadata as an audit signal rather than a hard policy decision. Finally, keep an eye on usage reports and the feedback loop from resolution reasons to gauge model accuracy and cost impact.
