GitHub Copilot Adds Model Selection Tiers, Sentry Workflow and Agent Updates

GitHub Copilot Adds Model Selection Tiers, Sentry Workflow and Agent Updates

AI-generated editorial illustration; not a documentary photograph.

GitHub has announced a set of Copilot updates spanning model selection, code review, production-error investigation and agent workflows. The September 14 weekly release also adds administrative tools for Business and Enterprise customers, while several VS Code 1.138 features are rolling out or remain in preview.

Three model selection tiers

Copilot's automatic model selection is gaining efficiency, balance and intelligence tiers, letting users prioritize cost, output quality or response time. GitHub says all three tiers choose from the same pool of available models. The feature is rolling out across VS Code, Copilot CLI and the Copilot app.

Code review gains validation tools

Copilot code review can now resolve feedback that has been addressed in later reviews while keeping unresolved comments open. It can also propose commit messages when users apply suggested changes. Reviews may use shell tools to validate work, and Lite reviews now consolidate findings produced by multiple agents.

Sentry errors move into Copilot

A new Sentry canvas in the Copilot app connects production crash reports with the coding workflow. Developers can inspect errors, stack traces and related context, then use Copilot to investigate a cause, validate a proposed fix and prepare a pull request.

Enterprise controls and VS Code agents

Copilot Business and Copilot Enterprise customers now have generally available usage metrics in the VS Code Agents window, including daily active users, sessions and message totals. These measurements are separate from editor-window metrics. Repository custom-property suggestions are in public preview, with organization and enterprise owners able to control them through policy.

Organizations using eligible usage-based billing can also handle requests for additional Copilot budget after users reach their AI credit limits. The generally available feature excludes enterprises with managed users. In VS Code 1.138, local Dev Container support for agents is rolling out gradually and requires Docker plus a supported configuration. Creating pull requests within the Agents window is supported, while automatic cleanup of inactive sessions is opt-in and in preview.

Source: GitHub Copilot weekly releases — September 14

Share this article
FacebookThreadsLinkedInX

Leave a Reply

Your email address will not be published. Required fields are marked *