Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →不慣れなコードベースを分析し、アーキテクチャマップ、主要なエントリポイント、規約、スターターCLAUDE.mdを含む構造化オンボーディングガイドを生成します。新しいプロジェクトに参加するか、リポでClaude Codeを初めてセットアップする場合に使用します。
.claude/skills/kunanonj-codebase-onboarding/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -27% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 100% | 0% |
| case-21 | ✗→✓ | ▲ Improved | 53% | 0% |
| case-01 | ✓→✗ | ▼ Worse | -70% | 0% |
| case-02 | ✓→✗ | ▼ Worse | -75% | 0% |
体系的に不慣れなコードベースを分析し、構造化オンボーディングガイドを作成。新しいプロジェクトに参加するか、既存リポでClaude Codeを初めてセットアップする開発者向けに設計。
すべてのファイルを読まずにプロジェクトについての生の信号を集めます。これらのチェックを並行して実行：
1. パッケージマニフェスト検出
→ package.json, go.mod, Cargo.toml, pyproject.toml, pom.xml
2. フレームワークフィンガープリント
→ next.config、nuxt.config、angular.json、vite.config
3. エントリポイント識別
→ main.*、index.*、app.*、server.*
4. ディレクトリ構造スナップショット
→ ディレクトリツリーの最上位2レベル
5. コンフィグとツール検出
→ .eslintrc、.prettierrc、tsconfig.json、Dockerfile
6. テスト構造検出
→ tests/、__tests__/、*.spec.ts、jest.config.*主要なモジュールとそれらの関係を特定します。
コード規約、命名パターン、プロジェクト固有のパターンを特定。
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-07 | fail→pass | 11,014 | 5,394 | -51% | 1 | 1 | 0% | 1,920 | 1,397 | -27% | 0 | 0 | — |
case-12 | fail→fail | 13,019 | 9,276 | -29% | 1 | 1 | 0% | 2,083 | 2,164 | +4% | 0 | 0 | — |
case-17 | fail→fail | 4,025 | 3,998 | -1% | 1 | 1 | 0% | 792 | 1,140 | +44% | 0 | 0 | — |
case-22 | pass→pass | 13,824 | 10,892 | -21% | 1 | 1 | 0% | 2,611 | 2,590 | -1% | 0 | 0 | — |
case-01 | pass→fail | 11,932 | 2,338 | -80% | 1 | 1 | 0% | 2,249 | 674 | -70% | 0 | 0 | — |
case-02 | pass→fail | 16,403 | 2,320 | -86% | 1 | 1 | 0% | 2,941 | 737 | -75% | 0 | 0 | — |
case-03 | fail→fail | 14,532 | 2,041 | -86% | 1 | 1 | 0% | 2,882 | 741 | -74% | 0 | 0 | — |
case-04 | pass→pass | 11,487 | 12,144 | +6% | 1 | 1 | 0% | 2,457 | 2,756 | +12% | 0 | 0 | — |
case-05 | pass→pass | 15,774 | 12,119 | -23% | 1 | 1 | 0% | 3,191 | 2,734 | -14% | 0 | 0 | — |
case-06 | pass→pass | 11,850 | 13,623 | +15% | 1 | 1 | 0% | 2,434 | 2,826 | +16% | 0 | 0 | — |
case-08 | pass→pass | 12,078 | 9,374 | -22% | 1 | 1 | 0% | 2,501 | 2,431 | -3% | 0 | 0 | — |
case-09 | pass→pass | 15,180 | 10,472 | -31% | 1 | 1 | 0% | 3,026 | 2,686 | -11% | 0 | 0 | — |
case-10 | pass→pass | 19,428 | 11,724 | -40% | 1 | 1 | 0% | 3,022 | 2,663 | -12% | 0 | 0 | — |
case-11 | pass→pass | 13,259 | 8,894 | -33% | 1 | 1 | 0% | 2,364 | 2,014 | -15% | 0 | 0 | — |
case-13 | fail→pass | 4,259 | 5,485 | +29% | 1 | 1 | 0% | 728 | 1,457 | +100% | 0 | 0 | — |
case-14 | pass→fail | 6,513 | 1,278 | -80% | 1 | 1 | 0% | 949 | 570 | -40% | 0 | 0 | — |
case-15 | fail→fail | 14,680 | 12,840 | -13% | 1 | 1 | 0% | 2,603 | 2,834 | +9% | 0 | 0 | — |
case-16 | pass→pass | 9,161 | 10,543 | +15% | 1 | 1 | 0% | 1,659 | 2,077 | +25% | 0 | 0 | — |
case-18 | pass→pass | 10,094 | 6,255 | -38% | 1 | 1 | 0% | 1,956 | 1,623 | -17% | 0 | 0 | — |
case-19 | pass→fail | 7,642 | 7,612 | -0% | 1 | 1 | 0% | 1,738 | 2,091 | +20% | 0 | 0 | — |
case-20 | pass→pass | 5,972 | 8,582 | +44% | 1 | 1 | 0% | 1,218 | 2,302 | +89% | 0 | 0 | — |
case-21 | fail→pass | 3,473 | 4,705 | +35% | 1 | 1 | 0% | 791 | 1,211 | +53% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of -22 percentage points is the difference between those two pass rates over the 22 comparable cases. 5 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.