GitHub Copilot Weekly Releases – Organizing functional elements from the official information of September 14

This article is a technical explanation and implementation example created using AI. Although the posted code and procedures are constructed based on primary information, actual device verification has not been performed by the author. Operations may vary depending on the environment and version.

In the GitHub Copilot weekly release (September 14), updates include the expansion of model selection options, code review feature updates, Sentry integration, administrative features, and the addition of agent features for VS Code. This article safely and practically interprets the published functional elements based on the official changelog, and organizes the specifications and configurations of each item in detail. Since actual devices have not been verified, actual screen displays and command operation results are not shown.

1. Purpose and Prerequisites of This Article

The purpose of this article is to accurately understand the numerous update details included in the latest GitHub Copilot weekly release information and to organize how they impact development environments and organizational management policies.

Prerequisites and Cautions

• This article is an explanation based on the official changelog, and actual device operation verification in individual development environments has not been performed.

• The rollout status of the described features varies depending on the plan being used (such as Business or Enterprise) and the version of each client (VS Code, Copilot CLI, Copilot app).

• Specific behaviors and display results in GUI or Windows environments are currently pending actual device verification.

2. GitHub Copilot Core Feature Updates

The primary information describes several important improvements regarding the basic functions of GitHub Copilot. These are deeply related to code reviews, model selection, and agent operations.

Smart Code Reviews

Copilot's code review features include the following updates:

• Automatic processing of resolved comments: In subsequent reviews, addressed comments are automatically resolved, leaving only unresolved feedback.

• Commit message suggestions: Appropriate commit messages are suggested when applying code proposals.

• Utilization of shell tools: Changes can be verified using shell tools during reviews.

• Integration of Lite reviews: Lite Review combines and presents findings from multiple agents.

Control over efficiency, quality, and latency in automatic model selection

The new automatic model selection feature allows users to choose their preferred balance between cost, quality, and response time.

• Three tiers: Available tiers include efficiency, balance, and intelligence.

• Common foundation: All three tiers select from the same available model pool.

• Deployment: This update is being rolled out progressively to VS Code, the Copilot CLI, and the Copilot app.

3. Business and enterprise features and administrator updates

Features related to usage measurement and budget management have been added for users and administrators on organizational plans (Business and Enterprise).

Usage measurement in the VS Code Agents window

The capability to measure usage and activity in a dedicated VS Code Agents window is now generally available (GA).

• Enterprise and organizational reports include daily active users (DAU), session counts, total user messages, and user-level activity data.

• These metrics are aggregated separately from traditional general editor window metrics.

Copilot suggestions for repository custom properties

Copilot suggestions are now available when choosing values for repository custom properties (Public Preview).

• For example, when creating a property indicating whether a repository should be public to the internet, Copilot suggests yes and no as permitted values.

• Organization and enterprise owners can enable or disable this suggestion feature via Copilot policies.

AI credit limits and budget requests

For organizations and enterprises, the ability to request a higher Copilot budget when reaching the AI credit limit is now generally available (GA) for Business and Enterprise plans billed based on usage.

• Owners and billing managers can approve, adjust, or deny requests from the settings screen.

• Once approved, access to AI credits is restored. Note that this feature is not available for enterprises with managed users.

4. GitHub Copilot App and Sentry integration

The GitHub Copilot app now includes crash-fixing capabilities leveraging reports from Sentry.

Fixing production crashes using the Sentry canvas

A new Sentry canvas is available to identify production crashes via Sentry reports and transition directly to code remediation.

• You can review errors, stack traces, and related context.

• You can collaborate with Copilot to investigate the cause, verify the fix, and prepare to create a pull request.

• To use this, you need to download the Copilot app from the official page.

5. Agent features in the VS Code 1.138 release

In connection with the VS Code release, updates have been made regarding agent execution environments and session management.

Using local Dev Containers

You can run local Dev Containers, including project tools and dependencies, within the Agents window to operate agents.

• Support is rolling out gradually, and requires both Docker and a supported Dev Container configuration.

Automatic session cleanup

A feature has been added to automatically Done mark inactive sessions where all pull requests have been merged (in preview).

• It also supports optional deletion after an individual grace period, and automatic cleanup is opt-in.

Creating Pull Requests from the Agents Window

You can now create pull requests directly from the agent host session without leaving the editor.

• You can review the generated title and description, select the draft status, and create the pull request directly or ask the agent to create it.

6. Organization of Processing Flow and Components

The following is a structural diagram showing the main functional areas described in the primary sources and their relationships.

flowchart TD
    A[GitHub Copilot Weekly Releases] --> B[Core Features]
    A --> C[Business & Enterprise]
    A --> D[Copilot App & Sentry]
    A --> E[VS Code 1.138 Updates]

    B --> B1[Code Review & Commit Messages]
    B --> B2[Auto Model Selection Tiers]

    C --> C1[VS Code Agents Usage Metrics]
    C --> C2[Repository Custom Property Suggestions]
    C --> C3[AI Credit Budget Requests]

    D --> D1[Sentry Canvas for Crash Fixes]

    E --> E1[Local Dev Containers in Agents]
    E --> E2[Inactive Session Auto-Cleanup]
    E --> E3[Direct PR Creation from Sessions]

7. Limitations and Cautions

• Lack of Physical Device Verification: Many of the features covered in this article are based on the official changelog at the time of writing, and physical verification in the authoring environment has not been performed.

• Phased Rollout: The timing of availability for new and preview features depends on the account plan, organization settings, and client version.

• Dependency on Management Policies: Features such as custom property suggestions and budget requests depend on organization owner or administrator settings.

8. Conclusion

We have organized the main updates in the GitHub Copilot weekly release (September 14).

Key Points to Check Before Execution

• Verify the versions of the plan you are using (Business / Enterprise) and the client (VS Code, CLI, App).

• For features controlled by organization policies (such as custom property suggestions and budget management), understand the configuration status of administrator permissions in advance.

• For features involving external tools, such as local Dev Containers or Sentry integration, verify that the necessary prerequisites (such as Docker installation) are met.

References

• GitHub Changelog: GitHub Copilot weekly releases — September 14