playbooks

home / skills / chachamaru127 / claude-code-harness / notebooklm

notebookLM skill

/skills/notebookLM

This skill helps you generate NotebookLM YAML or slide content from documents, enabling polished presentations and structured project documentation.

npx playbooks add skill chachamaru127/claude-code-harness --skill notebooklm

Review the files below or copy the command above to add this skill to your agents.

Files (3)
SKILL.md
3.8 KB
---
name: notebookLM
description: "NotebookLM用YAMLやスライドを生成。ドキュメント職人の腕の見せ所。Use when user mentions NotebookLM, YAML, slides, or presentations. Do NOT load for: implementation work, code fixes, reviews, or deployments."
description-en: "Generate NotebookLM YAML and slides. Document craftsman shows skill. Use when user mentions NotebookLM, YAML, slides, or presentations. Do NOT load for: implementation work, code fixes, reviews, or deployments."
description-ja: "NotebookLM用YAMLやスライドを生成。ドキュメント職人の腕の見せ所。Use when user mentions NotebookLM, YAML, slides, or presentations. Do NOT load for: implementation work, code fixes, reviews, or deployments."
allowed-tools: ["Read", "Write", "Edit"]
argument-hint: "[yaml|slides]"
---

# NotebookLM Skill

ドキュメント生成を担当するスキル群です。

## 機能詳細

| 機能 | 詳細 |
|------|------|
| **NotebookLM YAML** | See [references/notebooklm-yaml.md](references/notebooklm-yaml.md) |
| **スライド YAML** | See [references/notebooklm-slides.md](references/notebooklm-slides.md) |

## 実行手順

1. ユーザーのリクエストを分類
2. 上記の「機能詳細」から適切な参照ファイルを読む
3. その内容に従って生成

---

## 🔧 PDF ページ範囲読み取り（Claude Code 2.1.38+）

大型 PDF を効率的に扱うための機能です。

### ページ範囲指定で読み取り

```javascript
// ページ範囲指定で読み取り
Read({ file_path: "docs/spec.pdf", pages: "1-10" })

// 目次だけ確認
Read({ file_path: "docs/manual.pdf", pages: "1-3" })

// 特定のセクションのみ
Read({ file_path: "docs/api-reference.pdf", pages: "25-45" })
```

### ユースケース別の推奨アプローチ

| ケース | 推奨読み取り方法 | 理由 |
|--------|----------------|------|
| **100ページ超のPDF** | 目次(1-3) → 関連章のみ | トークン消費を最小化 |
| **仕様書レビュー** | セクション単位で範囲指定 | 必要な部分のみ精読 |
| **APIドキュメント** | エンドポイント一覧(目次)から開始 | 全体構造を把握してから詳細へ |
| **学術論文** | Abstract + 結論 → 本文 | 要点を先に把握 |
| **技術マニュアル** | 目次 + トラブルシューティング章 | 実用的な部分を優先 |

### NotebookLM YAML 生成時の活用例

```markdown
大型PDF（300ページの技術仕様書）からYAMLを生成する場合：

1. **目次を読む**（1-5ページ）
   Read({ file_path: "spec.pdf", pages: "1-5" })
   → 章立てを把握

2. **各章の冒頭を読む**（各章の最初の2ページ）
   Read({ file_path: "spec.pdf", pages: "10-11" })  // 第1章
   Read({ file_path: "spec.pdf", pages: "45-46" })  // 第2章
   → 各章の概要を把握

3. **重要セクションを精読**
   Read({ file_path: "spec.pdf", pages: "78-95" })  // APIリファレンス
   → 詳細な内容を抽出

この方法で、300ページすべてを読むことなく効率的にYAMLを生成できます。
```

### ベストプラクティス

| 原則 | 説明 |
|------|------|
| **段階的読み込み** | 目次 → 概要 → 詳細の順に読む |
| **関連ページのみ** | タスクに必要なページだけ指定 |
| **トークン節約** | 全ページ読み込みは最終手段 |
| **構造理解優先** | 目次で全体像を把握してから詳細へ |

### 従来の方法との比較

| 方法 | トークン消費 | 処理時間 | 精度 |
|------|------------|---------|------|
| **全ページ読み込み**（300ページ） | ~150,000 | 長い | 高 |
| **ページ範囲指定**（必要な30ページ） | ~15,000 | 短い | 高 |

→ **90%のトークン削減と処理時間短縮が可能**

Overview

This skill generates NotebookLM-compatible YAML and slide YAML to produce structured documents and presentations. It focuses on transforming user intents and source materials into well-formed NotebookLM artifacts. Use it when working on document drafting, presentation outlines, or YAML-driven NotebookLM workflows, but not for coding, implementation, or deployment tasks.

How this skill works

The skill classifies the user request to decide whether to produce NotebookLM YAML or slide YAML, then reads the appropriate reference templates and generation rules. For large PDFs it supports selective page-range reading to extract headings, summaries, and target sections. It outputs concise, NotebookLM-ready YAML or slide definitions aligned with the requested structure and reading strategy.

When to use it

  • When you need NotebookLM YAML to feed a NotebookLM document or knowledge base.
  • When creating slide YAML for automated slide generation or presentation outlines.
  • When summarizing large PDFs into structured YAML using page-range reads.
  • When you want stepwise extraction: table of contents → chapter intros → key sections.
  • Do NOT use for implementation work, code fixes, reviews, or deployments.

Best practices

  • Start with a clear request that indicates NotebookLM YAML or slides and desired output granularity.
  • Use page-range reading for large PDFs: begin with the table of contents, then target chapter starts and key sections.
  • Limit reads to relevant pages to save tokens and speed processing.
  • Provide contextual goals (audience, slide length, level of detail) to shape YAML structure.
  • Validate generated YAML in a test NotebookLM environment before wide use.

Example use cases

  • Generate NotebookLM YAML for a 300-page technical spec by reading the TOC and key API sections only.
  • Produce slide YAML for a 10-slide presentation from a product brief and target audience notes.
  • Extract chapter summaries from a long manual by specifying page ranges per chapter.
  • Create a concise NotebookLM knowledge card set from an academic paper (abstract + conclusions + select sections).

FAQ

How do page-range reads reduce token usage?

By reading only the table of contents and selected chapter pages you avoid processing the entire document, typically reducing token consumption by up to 90% compared to full-file reads.

Can this skill be used for code reviews or deployments?

No. This skill is tailored for document and slide generation and should not be used for implementation work, code fixes, reviews, or deployment tasks.