NotebookLM Skill

Provides full programmatic access to Google NotebookLM via the unofficial notebooklm-py library. Supports Python API and CLI — use the Python API for complex or multi-step workflows, CLI for quick one-off tasks.

Prerequisites

pip install "notebooklm-py[browser]"
playwright install chromium

# First-time auth (opens browser)
notebooklm login

Auth via env variable (CI/CD, no browser):

export NOTEBOOKLM_AUTH_JSON='{"cookies": [...]}'

Read references/auth.md for detailed authentication setup.

Quick Start Pattern

import asyncio
from notebooklm import NotebookLMClient

async def main():
    async with await NotebookLMClient.from_storage() as client:
        nb = await client.notebooks.create("My Research")
        await client.sources.add_url(nb.id, "https://example.com/article", wait=True)
        result = await client.chat.ask(nb.id, "Summarize the main points")
        print(result.answer)

asyncio.run(main())

API Reference Summary

Client Structure

NotebookLMClient
├── .notebooks   → NotebooksAPI
├── .sources     → SourcesAPI
├── .artifacts   → ArtifactsAPI
├── .chat        → ChatAPI
├── .research    → ResearchAPI
├── .notes       → NotesAPI
└── .sharing     → SharingAPI

Notebooks (client.notebooks)

Sources (client.sources)

Chat (client.chat)

Artifacts — Generate (client.artifacts)

All generate methods return a GenerationStatus with taskid. Use waitfor_completion() to poll until done.

Wait pattern:

status = await client.artifacts.generate_audio(nb_id, instructions="Focus on key facts")
final = await client.artifacts.wait_for_completion(nb_id, status.task_id, timeout=300)
if final.is_complete:
    await client.artifacts.download_audio(nb_id, "./podcast.mp3")

Research (client.research)

# Start web research
research = await client.research.start(nb_id, "quantum computing trends", source="web", mode="deep")
# Poll until complete
status = await client.research.poll(nb_id)
# Import discovered sources
await client.research.import_sources(nb_id, task_id, sources[:10])

CLI Quick Reference

# Session
notebooklm login
notebooklm use <notebook_id>
notebooklm status

# Notebooks
notebooklm list
notebooklm create "My Notebook"
notebooklm delete <id>
notebooklm rename "New Title"

# Sources
notebooklm source add "https://example.com"
notebooklm source add ./paper.pdf
notebooklm source add-research "AI trends 2025" --mode deep --import-all
notebooklm source list

# Chat
notebooklm ask "What are the key themes?"

# Generate (--wait blocks until done)
notebooklm generate audio "make it engaging" --wait
notebooklm generate video --style whiteboard --wait
notebooklm generate quiz --difficulty hard --wait
notebooklm generate flashcards --quantity more --wait
notebooklm generate slide-deck --wait
notebooklm generate infographic --orientation portrait --wait
notebooklm generate report --format study-guide --wait
notebooklm generate mind-map

# Download
notebooklm download audio ./podcast.mp3
notebooklm download video ./overview.mp4
notebooklm download quiz --format json ./quiz.json
notebooklm download flashcards --format markdown ./cards.md
notebooklm download slide-deck ./slides.pdf
notebooklm download mind-map ./mindmap.json
notebooklm download data-table ./data.csv

Common Workflows

Research → Podcast

See references/workflows.md for the full pattern.

Bulk Source Import

urls = ["https://...", "https://...", ...]
for url in urls:
    await client.sources.add_url(nb_id, url, wait=False)
    await asyncio.sleep(1)  # Rate limiting

Quiz Generation + JSON Export

status = await client.artifacts.generate_quiz(nb_id, difficulty="hard")
final = await client.artifacts.wait_for_completion(nb_id, status.task_id)
await client.artifacts.download_quiz(nb_id, "quiz.json", output_format="json")

Error Handling

from notebooklm import RPCError

try:
    result = await client.notebooks.create("Test")
except RPCError as e:
    # Session expired → re-run `notebooklm login`
    # Rate limited → wait and retry
    print(f"Error: {e}")

Common issues:

• RPCError auth failure → re-run notebooklm login or call await client.refresh_auth()
• Rate limits → add await asyncio.sleep(2) between bulk operations
• Source not ready → use wait=True in addurl() / addfile()

Reference Files

• references/auth.md — Detailed auth setup, CI/CD, environment variables
• references/workflows.md — Complete workflow examples (research→podcast, bulk import, etc.)
• references/artifact-options.md — All generation options, formats, and styles