Interpreting the code quality results for your repository

View GitHub Code Quality findings for your default branch.

この機能を使用できるユーザーについて

書き込み アクセスを持つユーザー

GitHub Code Quality は、 GitHub Team または GitHub Enterprise Cloud 上の組織所有リポジトリで使用できます

この記事の内容

メモ

GitHub Code Quality は現在 パブリック プレビュー にあり、変更される可能性があります。 パブリック プレビュー の間、Code Quality は課金されませんが、Code Quality スキャンでは GitHub Actions 分が消費されます。

Prerequisites

Viewing the full backlog of code quality results

  1. リポジトリの [セキュリティ] タブに移動します。
  2. クリックして コードの品質 を展開し、標準の結果 をクリックします。

Alternatively, if you want to view AI-powered findings for the most recently changed files, see Improving the quality of recently merged code with AI.

Exploring the backlog for your repository

The "標準の結果" dashboard shows all the results found by CodeQL analysis on the default branch of your repository. This view helps you visualize the full backlog of quality results and prioritize work to fix specific types of problems.

The overview, at the top of the page, summarizes the maintainability and reliability of the codebase.

Screenshot of the "標準の結果" dashboard for code quality results. The summary is outlined in dark orange.

Underneath the overview, the full list of results is shown with a header with filters that you can use to focus on a specific set of findings. The results are:

  • Grouped by the rule that detected each finding
  • Within each rule, ordered by file path alphabetically

Explore the results by expanding a rule to list the affected files and clicking on the name of a rule to see full details of the findings.

Screenshot of the Rules table on the "標準の結果" dashboard for code quality. The "Overwritten property" rule name is outlined in dark orange.

Interpreting ratings and metrics

Code quality results should always be interpreted in the context of your repository. For example:

  • Small repositories, or repositories with only a small amount of code written in supported languages, tend to have few results and good ratings.
  • Repositories with a lot of generated code may have many maintenance results, lowering the rating for maintainability. This is not a problem if the source code itself is maintainable.
  • Large repositories with a lot of code in a fully supported language often have many results even if the majority of the code has good maintainability and reliability standards.

To learn more about the metrics and how the ratings are calculated, see メトリックと評価のリファレンス.

Next steps