GitHub Copilot code review adds automatic comment closing and ensemble reviews in Lite mode
GitHub Copilot code review now automatically closes resolved comments, generates smart commit messages and uses an ensemble of multiple agents in Lite mode — with higher-quality findings and lower costs, according to the company.
GitHub has updated Copilot code review with several new features. The tool now closes its own comments when a user submits a commit addressing the feedback in question, so open threads correspond only to feedback that still requires attention. When a suggestion from Copilot is applied, the tool generates its own smart suggestion describing the specific change made instead of a standard commit message.
According to GitHub, the review agent now uses a broader set of shell tools from Copilot SDK, running behind the so-called Copilot agent firewall — allowing it to run build commands, tests and targeted scripts, and obtain information from available tools and APIs to verify the code under review. The company states that, according to its experiments, this change led to more positive developer feedback on comments from Copilot and the detection of more severe findings with fewer minor comments (“nits").
According to GitHub, Lite effort level now uses an ensemble of multiple agents instead of a single agent — each agent contributes its own perspective on the code, and Copilot combines the results into a single review. According to company experiments, this approach increased the average number of addressed comments by 47 % for high-severity findings, by 31 % for medium and by 11 % for low, while reducing review costs by approximately 8 %.
Why it matters
Automatically closing resolved comments and generating commit messages reduces the number of tasks a developer must perform manually during code review. According to GitHub, the ensemble approach in the cheaper Lite mode also improves the detection of severe issues at lower costs, which is relevant for teams considering how many code reviews to automate using an AI tool and at what price.
Two audiences, two different impacts
What this means
For individuals
Developers using Copilot code review no longer need to manually close resolved comments, and when they apply a suggestion, the tool automatically offers a commit message that matches the specific change, speeding up the routine code review workflow.
For a business
According to GitHub, the ensemble approach in Lite mode increases addressed high-severity findings by 47 %, medium by 31 % and low by 11 %, while saving approximately 8 % on review costs — relevant for companies planning the costs of using Copilot code review in development teams.
DevelopmentCheck the original
Event sources
clearly official source · 1 publisher, 0 independent. We count feeds from the same owner only once.