プロジェクトガイドラインへの準拠をチェックし、バグ、スタイル違反、コード品質問題を検出する。PRレビュー時に使用。
comment-analysis
コードコメントの正確性、完全性、長期的な保守性を分析する。JSDoc、コメント、README等のドキュメントが変更された場合に使用。
インストール方法を見る含まれるファイル(1)
- SKILL.md2.0 KB
SKILL.md(原文)
インストールする前に、エージェントに与えられる指示の中身を確認できます。
Comment Analysis スキル
目的
コードコメントの正確性、完全性、長期的な保守性を分析し、技術的負債を防ぐ。
検証プロセス
1. 事実精度の確認
- 関数署名(パラメータ、戻り値の型)が一致しているか
- 説明された動作が実際のコードと一致しているか
- 参照されている型、関数、変数が存在するか
2. 完全性の評価
- 重要な仮定がドキュメント化されているか
- 副作用が記載されているか
- エラー状態が説明されているか
- 複雑なアルゴリズムの解説があるか
3. 長期的価値の評価
- 「なぜ」を説明しているか(「何を」だけでなく)
- 非自明なロジックを説明しているか
- 過度に冗長でないか
4. 誤解要素の特定
- 曖昧な表現の有無
- 古い参照の有無
- 時代遅れな仮定の有無
出力形式
# Comment Analysis Results
## Critical Issues(事実として不正確)
### [ファイル名:行番号]
- コメント: "..."
- 問題: コメントが実際のコードと矛盾している
- 実際のコード: ...
- 推奨修正: ...
## Improvement Opportunities(改善可能)
### [ファイル名:行番号]
- コメント: "..."
- 問題: 情報が不完全
- 推奨追加内容: ...
## Positive Findings(良い例)
### [ファイル名:行番号]
- コメント: "..."
- 評価: 「なぜ」を適切に説明している
## Summary
- Critical Issues: X
- Improvement Opportunities: X
- Positive Findings: X
注意事項
- 変更されたファイルのコメントのみを分析
- 事実の不正確さを最優先で報告
- 良い例も報告して学習を促進
- 主観的な「スタイル」の指摘は避ける
レビュー
まだレビューはありません。使ってみた感想をお寄せください。
同じリポジトリのスキル
概要と使いどころ
Enforces project coding standards during code review: naming conventions (PascalCase for classes, camelCase for methods/variables, UPPER_SNAKE_CASE for constants, kebab-case for files), standard file structure, anti-magic-number policy, file size limits (500/800 lines), security patterns, and design patterns. Use when reviewing code, writing new code, or checking adherence to project conventions.
日本語の概要は準備中です。原文の説明を表示しています。
機能を保持したまま不要な複雑性を排除し、コードの簡潔性と可読性を向上させる。30行超の関数や3段以上のネスト検出時に使用。
サイレント失敗を検出し、堅牢なエラーハンドリングを確保する。PRレビューでのエラーハンドリング検査時に使用。
Enforces error handling standards: silent error prohibition, custom error class hierarchy (AppError base with category never-fallback / transient / permanent, cause preservation, ValidationError / UnauthorizedError / UpstreamError etc. — canonical definitions in PATTERNS.md), external-boundary normalization (normalizeExternalError), Result pattern (Result.ok/Result.fail), proper try-catch with error type checking, structured error logging with a single Logger contract, fallback prohibition categories (isNeverFallback), and HTTP status code mapping. Use when implementing error handling, reviewing catch blocks, or designing error responses.
日本語の概要は準備中です。原文の説明を表示しています。
Creates and reviews agent skills safely by enforcing deletion-prevention harness rules: no fixed absolute temp paths, no cwd-dependent cleanup, no wildcard deletes, no workspace-external file operations, no silent cleanup failures, and mandatory temp-root validation before any destructive command. Use when creating a new skill, updating an existing SKILL.md, or reviewing prompts/runbooks that include temp files, extraction, cleanup, PowerShell, bash, rm, del, Remove-Item, or archive expansion.
日本語の概要は準備中です。原文の説明を表示しています。