GitHub Expands Copilot Code Review with 'Max' Effort Level and Automated Approvals

Enhanced AI Code Analysis

GitHub has announced the release of a new "Max" review effort level for its Copilot code review tool, now available in public preview. This feature aims to provide more comprehensive analysis for high-risk pull requests, such as updates to authentication logic or core shared libraries.

Unlike previous levels like "Lite" or "Balanced," the Max setting employs multiple independent AI agents across different models. An aggregation agent is used to reconcile findings and address potential discrepancies before finalizing the feedback, a process designed to identify issues that a single model might overlook.

General Availability of Copilot Approvals

Alongside the new review level, GitHub has moved the Copilot approvals feature to general availability. Key aspects of this feature include:

• Approval Integration: Copilot can now formally approve a pull request, with its approval counting toward the repository's mandatory requirements.
• Lifecycle Management: If new commits are added to a branch after Copilot provides an approval, the system automatically dismisses it, requiring a new review cycle to ensure accuracy.
• Configurability: Repository administrators can restrict which file paths Copilot is authorized to approve.

Implementation and Access

These updates are available to users on Copilot Pro, Pro+, Max, Business, and Enterprise plans. Organizations and administrators have control over the implementation:

• Review Settings: Admins can set maximum effort levels for repositories or organizations, preventing the use of "Max" if desired.
• Approval Controls: Approvals remain off by default and must be explicitly enabled by administrators at the desired organizational or repository level.
• Default Effort: Developers can select the review effort level per pull request or configure a default setting for their projects.

GitHub noted that because the Max effort level involves more extensive processing, it may require more time and consume more AI credits compared to lower effort tiers.

How to Verify

Users interested in implementing these features can manage their settings by visiting the official GitHub documentation on "Configuring code review by GitHub Copilot." Further updates and community discussions are hosted on the GitHub Copilot community forums.