Service eligibility and regional restrictions
PuppyIP serves only compliant overseas businesses and their authorized personnel. Proxy services are not available in mainland China. The service may only be used for lawful business activities outside mainland China. Use of this service within mainland China is prohibited.
Hosting a proxy IP or server overseas does not change these restrictions. The service must not be provided to end users in mainland China through relaying, forwarding, sharing or resale. Before use, read the Terms of Service.
Key Takeaways
- Adaptive, cross-repository context and review secret detection are public previews for Copilot Pro, Pro+, Max, Business and Enterprise. This does not verify access in every account.
- Adaptive chooses Lite, Balanced or Max and shows the actual level. Balanced remains the built-in default, and maximum effort limits still apply.
- Cross-repository context currently covers public repositories, including related public code, documentation, custom instructions and skills. Private cross-repository access is not promised.
- Secret detection flags possible credentials in PR review comments. It differs from standalone secret scanning in history coverage, alerts and eligibility; a confirmed exposed credential still needs rotation or revocation.
Separate the three previews from the existing default
GitHub’s November 17, 2026 announcement identifies Adaptive, context from public repositories and secret detection during reviews as public previews. Eligible plans are Copilot Pro, Pro+, Max, Business and Enterprise. The announcement does not list Free or give a completion time for rollout to every account or region.
Adaptive selects review effort automatically. It evaluates a pull request’s complexity and risk, then chooses Lite, Balanced or Max. Each review displays the chosen level. This release does not change the built-in default from Balanced to Adaptive.
Select Adaptive for one PR or for future reviews
For a single PR, select Adaptive from its Reviewers bar. To make it your personal default, open Copilot settings from your profile-picture menu, go to Code review and choose Adaptive under Review effort level. Current documentation also supports defaults at repository, organization and enterprise levels.
Choose a one-off review or a default change within your team’s authorization rules. Your personal effort selection applies to automatic and manually requested reviews. Turning off Automatic Copilot code review does not clear the effort selection or stop it applying to manual reviews. Review depth and automatic triggering are separate settings.
Repository administrators can inspect Review effort level and Maximum review effort under Settings → Copilot → Code review. Repositories cannot raise an enterprise or organization maximum, and Adaptive never exceeds the applicable cap. If the maximum is Balanced, do not expect a risky PR to make Adaptive select Max.
What can cross-repository context read?
This preview covers public repositories. During a review, Copilot can find relevant public code dependencies, documentation, custom instructions and skills to understand how a change relates to other code. It does not grant new access to every private repository.
Current configuration documentation provides Search public repositories under repository Settings → Copilot → Code review. It is on by default; administrators can turn it off to prevent public repository searches during reviews. Check the search policy, supplied instructions and context needed for the PR separately.
For example, suppose one public library changes an interface used by another public library. Cross-repository information may help review the dependency impact. This is an illustrative use case, not a review performed for this guide or proof that Copilot will find every caller. Teams must still check private dependencies through their existing access and review processes.
What should you do after a secret-detection comment?
The preview flags potentially exposed credentials in PR review comments. GitHub describes a combination of deterministic known-format detectors and AI/ML detection. The former uses work with more than 150 technical partners and covers over 500 secret types; the latter can identify secrets without recognizable formats. Those counts do not guarantee detection of every secret.
Enable secret detection under Settings → Copilot → Code review at repository or organization level, then have an authorized maintainer inspect the resulting comments. Do not commit a real credential merely to test the feature, or copy credentials into public discussions or articles. An absence of comments is not proof that code contains no secrets.
If a credential has been exposed, promptly rotate or revoke it through your team’s incident process and check the application configuration after replacement. Removing a line from the PR, accepting an AI suggestion or resolving a comment does not establish that the old credential is invalid. GitHub’s standalone secret scanning documentation also calls for immediate rotation after exposure.
How does this relate to secret scanning and push protection?
The announced feature points out secrets in review comments. Standalone secret scanning scans the full Git history across branches and produces alerts on the relevant security page. A review comment does not establish that history was fully scanned, a push was blocked or a credential was automatically revoked. Each outcome needs its own evidence.
Standalone secret scanning has separate eligibility. Current documentation says it runs automatically for free on public repositories. Organization-owned private and internal repositories require GitHub Secret Protection enabled on GitHub Team or Enterprise Cloud. That rule does not establish that the Copilot preview is free for every private repository.
Evaluate the result and costs before broader use
For an appropriate, authorized trial PR, record the selected mode, applicable maximum and actual effort displayed in the review. Have a maintainer assess whether comments are relevant and suggestions are correct. To examine how review feedback is handled, open Insights → Repository insights in a repository you can read. Access to statistics alone does not demonstrate review quality or cost savings.
Current configuration documentation says Balanced and Max use more AI credits and may consume marginally more GitHub Actions minutes. Adaptive’s selected effort is capped, but that is not a spending cap. Check the existing payer, allowance and budgets separately. The reviewed sources do not establish one price for all PRs, a complete regional eligibility table or a savings percentage; public preview does not mean free.
This guide reviews public announcements and documentation. It did not enable organization features, submit a PR, handle real secrets or measure review results. If an option is missing, first check the plan, administrator policies, repository scope and current preview status. A proxy exit cannot add subscription entitlements or bypass these conditions.
Sources
Frequently Asked Questions
Is Copilot code review for Azure Repos still in preview?
The November 17 announcement separately makes Copilot code review for Azure Repos generally available. Usage is billed in GitHub AI credits to the Azure subscription linked to the Azure DevOps organization. The announcement lists Lite or Balanced as default effort choices there. It does not establish that Adaptive, public cross-repository context or the full GitHub plan conditions in this guide also apply to Azure Repos. Check its dedicated Azure DevOps documentation.
Will an explicit Lite selection automatically change to Adaptive?
This announcement does not say explicit Lite selections will change to Adaptive, and it confirms that Balanced remains the default effort. Inspect the actual personal, repository, organization or enterprise settings and applicable maximum. Select Adaptive when you want it; a feature announcement is not evidence that an existing configuration changed.