GitHub Copilot Code Review: Severity Labels and Grouped Comments
GitHub CopilotGitHub has improved the comment experience for Copilot Code Review, adding severity labels (High, Medium, or Low) to each suggestion so developers can instantly prioritize what to fix first. Additionally, grouped comments consolidate repetitive feedback — if the same issue appears across multiple code locations, Copilot surfaces it only once, reducing noise especially on larger pull requests. Both improvements are available to users enrolled in the new pull request experience.
Sources & Mentions
3 external resources covering this update
Smarter, Quieter Code Review Feedback
GitHub Copilot Code Review received a meaningful upgrade to how it surfaces feedback on pull requests, addressing two of the most common friction points developers encounter: information overload and unclear prioritization.
Severity Labels for Every Comment
Each Copilot code review comment now carries a severity label — High, Medium, or Low — displayed in the top-right corner of the comment card. This classification gives developers an at-a-glance signal for how urgently a suggestion should be addressed.
Before this change, all Copilot suggestions were presented with equal weight, leaving developers to manually judge whether a flagged issue was a critical security hole or a minor style inconsistency. With severity labels, the triage work is done upfront: a High-severity comment demands immediate attention, while a Low-severity suggestion can be safely deferred or addressed in a follow-up pass.
This is particularly valuable on large pull requests where Copilot may surface dozens of comments. Developers can now scan the severity column first, resolve the critical issues, and make an informed decision about what to leave for later.
Grouped Comments to Eliminate Repetition
The second improvement targets a frustrating pattern that emerges in larger codebases: Copilot flagging the same problem multiple times across different locations in the same PR. If a naming convention issue or a missing null check appears in ten places, previous behavior would generate ten separate comments — creating a wall of noise that obscures more important, unique feedback.
With grouped comments, Copilot consolidates similar observations and surfaces the issue once. This keeps the review thread focused on distinct, actionable items rather than repeated variations of the same suggestion.
Availability
Both features are available to users opted into the new pull request experience on GitHub. The changes complement the agentic code review architecture GitHub shipped in March 2026, which added broader repository context gathering to improve the quality and relevance of suggestions.