NotebookLM Skill
Manage NotebookLM notebooks via the notebooklm CLI.
Setup
# Install notebooklm CLI
pip install notebooklm
# Authenticate (opens browser)
notebooklm login
Core Actions
List all notebooks
notebooklm list
Create a new notebook
notebooklm create "Notebook Title"
Returns the notebook ID (e.g. 9e2f71b6-373e-432f-9ed0-e6a017cfdcac).
Add a URL or text source to a notebook
# Set notebook as current context
notebooklm use <notebook_id>
# Add URL source
notebooklm source add "https://example.com/article"
# Add multiple sources (wait for each)
notebooklm source add "https://url1.com"
notebooklm source wait <source_id>
Ask a question (chat with notebook)
# Ask in current notebook (set with notebooklm use <id>)
notebooklm ask "What are the main topics?"
notebooklm ask "Summarize this content in Chinese"
notebooklm ask "Generate 5 quiz questions" --json
# Start new conversation
notebooklm ask "Summarize" --new
Get notebook summary
notebooklm summary
Generate learning materials
notebooklm generate slide-deck "create a presentation about X"
notebooklm generate quiz "create a 10-question quiz"
notebooklm generate flashcards "make flashcards for key terms"
notebooklm generate infographic "visual summary"
notebooklm generate report "detailed report"
Poll artifact status and download
# Poll (after generate, check if ready)
notebooklm artifact poll <artifact_id>
# List artifacts
notebooklm artifact list
# Download slide deck as PPTX/PDF
notebooklm download slide-deck --artifact <id> --format pptx
notebooklm download slide-deck --artifact <id> --format pdf
# Download as PDF
notebooklm download slide-deck
# Download quiz/flashcards
notebooklm download quiz
notebooklm download flashcards
Intent → Action Mapping
| User says | Action |
|---|---|
| "列出我的 notebook" / "list notebooks" | notebooklm list |
| "新建 notebook 叫 X" | notebooklm create "X" |
| "在 notebook 里加入这个链接" | notebooklm use <id> && notebooklm source add <url> |
| "导入这段文字到 notebook" | notebooklm use <id> && notebooklm source add "<text>" |
| "总结这个 notebook" | notebooklm use <id> && notebooklm summary |
| "问我这个 notebook X 问题" | notebooklm use <id> && notebooklm ask "X" |
| "生成 PPT" / "生成幻灯片" | notebooklm use <id> && notebooklm generate slide-deck then poll & download |
| "生成测验" / "生成 quiz" | notebooklm use <id> && notebooklm generate quiz |
| "生成闪卡" | notebooklm use <id> && notebooklm generate flashcards |
| "下载生成的 PPT" | notebooklm download slide-deck --format pptx |
Workflow: Create notebook + add sources + generate PPT
# 1. Create
notebooklm create "English Grammar Notes"
# 2. Set as current
notebooklm use <new_id>
# 3. Add sources
notebooklm source add "https://example.com/grammar-guide"
# wait for ready
notebooklm source wait <source_id>
# 4. Generate slide deck
notebooklm generate slide-deck "Create educational slides about modal verbs"
# 5. Poll until done (takes 2-5 min)
notebooklm artifact poll <artifact_id>
# 6. Download
notebooklm download slide-deck --artifact <artifact_id> --format pptx
Notes
- Artifact generation takes 2-10 minutes — poll periodically with
notebooklm artifact poll <id> - Use
--jsonflag withaskfor machine-readable responses - Use
notebooklm statusto check current notebook context - Use
notebooklm source listto see all sources in a notebook - Artifact types available: slide-deck, quiz, flashcards, infographic, report, audio, video, mind-map, data-table
