GitHub Copilot code review has been expanded with three public‑preview features—adaptive review effort, cross‑repository context, and secret detection—and the Azure DevOps integration is now generally available. For engineers who rely on AI‑assisted reviews, the updates promise tighter alignment between review depth and code risk, broader visibility into dependent code, and early warning of exposed credentials, all of which affect pipeline design and operational monitoring.
Adaptive review effort
Copilot now evaluates each pull request’s complexity and risk to select one of three effort levels: Lite, Balanced, or Max. The selection is displayed in the review output, so teams can see which tier was applied. Users can enable Adaptive per pull request via the Reviewers bar, or set it as the default at personal, repository, organization, or enterprise scope. The default remains Balanced, and Adaptive will never exceed the maximum effort level configured for the surrounding scope. This reduces the need for manual effort‑level decisions while ensuring that high‑risk changes receive a deeper analysis.
Cross‑repository context
When enabled, Copilot code review can read related public repositories during analysis. This allows the reviewer to surface dependencies that live outside the current repo, locate documentation stored elsewhere, and apply custom instructions or skills defined in other projects. The feature is limited to public repositories in the preview stage, so private cross‑repo data remains untouched. Practically, this means a change that touches a shared library or an API contract can be evaluated against its real consumers, potentially catching integration regressions earlier.
Secret detection in reviews
Copilot now flags exposed secrets directly in pull‑request comments. Detection combines deterministic regex‑based scanners covering 500+ known secret formats—built through GitHub’s partnership program—with an AI/ML model that identifies secrets lacking a recognizable pattern. Enabling the feature requires toggling a setting under the repository or organization’s Copilot → Code review options. Once active, reviewers will see comments that point out the exact line and type of secret, giving engineers a chance to remediate before merge.
Azure DevOps GA and repository insights
The Azure Repos integration, previously in preview, is now generally available. Administrators can turn on Copilot code review at organization, project, or repository level, add branch policies to request reviews automatically, and choose a default effort level. Custom instructions and agent skills remain configurable, and suggestions can be applied and committed in a single step. Billing for Azure DevOps usage is recorded as GitHub AI credits against the linked Azure subscription, enabling cost tracking via Azure Cost Management. Additionally, Copilot now populates the Insights tab with per‑repository review outcome metrics, allowing teams to measure adoption and effectiveness.
Related CloudNinjas coverage: AI engineering.
What This Means For Practitioners
Evaluate your current review policies and consider setting Adaptive as the default to reduce manual effort‑level selection. Verify that the maximum effort level configured at the enterprise or organization tier aligns with your risk tolerance. If you rely on shared libraries, enable cross‑repository context for public repos to surface downstream impact. Turn on secret detection and incorporate the generated comments into your pre‑merge gate to avoid credential leakage. Finally, monitor the new Insights data and Azure AI credit usage to ensure the feature’s cost and benefit balance meets your operational goals.


