notebooklm

name	notebooklm
description	Automate Google NotebookLM via the notebooklm-py CLI and Python API — create notebooks, add sources (URLs, PDFs, YouTube, Drive), ask questions, generate podcasts/videos/quizzes/reports, and download artifacts. Use this skill whenever the user mentions NotebookLM, wants to create audio overviews or podcasts from documents, generate study materials (quizzes, flashcards), produce video explainers, build mind maps or infographics from research, or needs to programmatically interact with Google NotebookLM in any way. Also use when the user asks about setting up notebooklm-py, authenticating with NotebookLM, or integrating NotebookLM into scripts and CI/CD pipelines.
allowed-tools	Read, Write, Bash

NotebookLM CLI & Python API

Automate Google NotebookLM through the notebooklm-py package (unofficial community client; CLI + async Python API). This file is the workflow and orientation; full command syntax and recipes are in the references.

• references/cli-reference.md — complete CLI commands, flags, advanced features.
• references/python-api.md — full async Python API reference.
• references/setup-guide.md — install, auth, multiple accounts, CI/CD.
• references/troubleshooting.md — detailed troubleshooting.

Note: uses undocumented Google APIs — community project, not affiliated with Google; rate limits apply.

Quick Setup

pip install "notebooklm-py[browser]" && playwright install chromium
notebooklm login          # opens browser for Google login
notebooklm auth check --test   # verify environment

Platform-specific setup, multiple accounts, and CI/CD config: references/setup-guide.md.

Core Workflow

Five steps — one example each; full flags in references/cli-reference.md.

1. Create — notebooklm create "My Research Project".
2. Add sources — notebooklm source add "<url | ./file.pdf | youtube-url>"; source add-drive, source add-research "<query>" for AI source discovery. Accepts PDF/Word/Markdown/text/audio/video/ images/Drive/YouTube/pasted text. Per-plan source limits: Standard 50, Plus 100, Pro 300, Ultra 600. Indexing takes time — notebooklm source wait <id> before asking/generating.
3. Ask — notebooklm ask "What are the key findings?" (add --save-as-note to keep the answer).
4. Generate an artifact (add --wait to block until ready, -s <source_id> to limit to sources):
   • audio (podcast), video / cinematic-video, quiz, flashcards, slide-deck, infographic, mind-map, report, data-table.
   • e.g. notebooklm generate audio "focus on practical implications" --wait.
5. Download — notebooklm download <artifact> <path> (e.g. download audio ./podcast.mp3); options --all, --name "<fuzzy>", -a <artifact_id>, --format <fmt>, --dry-run.

Generation Timeframes

Artifact	Typical Duration
Mind maps	Instant (synchronous)
Quizzes, flashcards	1–5 minutes
Audio podcasts	5–15 minutes
Reports, slide decks	5–15 minutes
Video explainers	10–30 minutes
Cinematic videos (Veo 3)	30–40 minutes
Source indexing	30 s – 10 minutes

For long operations use --wait, or poll with notebooklm artifact list / artifact poll <task_id>.

Artifact Options

• Audio — --format deep-dive|brief|critique|debate, --length short|default|long, 80+ languages (notebooklm language set pl).
• Video — --format explainer|brief|cinematic (Veo 3), --style auto|classic|whiteboard|kawaii|anime| watercolor|retro-print|heritage|paper-craft.

Python API

For programmatic/CI workflows use the async client (NotebookLMClient.from_storage() → notebooks.create, sources.add_url, chat.ask, artifacts.generate_audio / wait_for_completion / download_audio). Full reference: references/python-api.md.

Troubleshooting

• Auth errors / CSRF — notebooklm auth check --test to diagnose, then notebooklm login to re-auth.
• Rate limiting — wait 5–10 min or use --retry <n> for automatic backoff.
• Generation stuck — use --wait or poll artifact list / artifact poll <task_id>.

More: references/troubleshooting.md.

Key Tips

• Use --wait when you need to download immediately after generating; --json for machine-parseable output.
• For parallel/agent workflows pass explicit notebook IDs with -n instead of relying on context.
• Use profiles (-p work) for multiple Google accounts; -s <source_id> to scope generate/ask.
• source fulltext retrieves indexed content (not in the web UI).
• Slide decks revise per-slide: notebooklm generate revise-slide "prompt" --artifact <id> --slide N.
• Export artifacts to Google Docs/Sheets: notebooklm artifact export <id> --type docs.