notebooklm

name	notebooklm
description	Interact with Google NotebookLM notebooks — chat with the AI, generate artifacts (slides, audio, video, mind maps, quizzes, flashcards, infographics, reports, data tables), manage sources (add URLs, YouTube, files, text), run research (fast/deep web research), and manage notes. Use when the user wants to query, create content from, or manage their NotebookLM notebooks and sources.

NotebookLM AI Plugin

Supports:

• Chat with Notebook AI (source-grounded Q&A with citations)
• Slide Deck generation (PDF/PPTX)
• Audio Overview (M4A -- deep dive, brief, critique, debate formats)
• Video Overview (MP4 -- classic, whiteboard, kawaii, anime, watercolor styles)
• Mind Map (HTML)
• Flashcards (HTML/JSON)
• Quiz (HTML/JSON)
• Infographic (PNG -- landscape, portrait, square)
• Report (Markdown -- briefing doc, study guide, blog post)
• Data Table (CSV / Google Sheets export)
• Source management (add URLs, YouTube, files, pasted text; list, delete)
• Fast/Deep web research with auto-import
• Notes management (create, update, delete, list)
• Notebook library management (add, list, search, activate)

Quick start

# Authenticate first (opens Chrome for Google login)
npx -y bun scripts/main.ts login

# Add a notebook to library
npx -y bun scripts/main.ts notebooks add https://notebooklm.google.com/notebook/abc123 --name "My Research"

# Generate artifacts
npx -y bun scripts/main.ts generate slide_deck --notebook abc123 --output slides.pdf
npx -y bun scripts/main.ts generate audio --format deep_dive --length long
npx -y bun scripts/main.ts generate video --style whiteboard --output explainer.mp4
npx -y bun scripts/main.ts generate quiz --difficulty medium --quantity more --json
npx -y bun scripts/main.ts generate mind_map --output mindmap.png
npx -y bun scripts/main.ts generate infographic --orientation portrait --output info.png
npx -y bun scripts/main.ts generate report --format study_guide --output report.md
npx -y bun scripts/main.ts generate flashcards --difficulty easy --json
npx -y bun scripts/main.ts generate data_table --output data.csv

Commands

Authentication

npx -y bun scripts/main.ts login

First run opens Chrome for Google login. Cookies are cached for subsequent runs.

# Force cookie refresh
npx -y bun scripts/main.ts login --force

GARY — the candidate context notebook (create / RECREATE)

GARY keeps ONE notebook per user = the candidate RAG (CV + technical/soft skills + role families + onboarding answers). It is created during onboarding. If the user deletes it from the NotebookLM website (or the saved id 404s), recreate it the SAME way — this is the supported recovery path, done by you (the LM supervisor):

# 1) (Re)create the canonical notebook — payload is [title]; the CLI extracts the real UUID (response index [2]).
npx -y bun scripts/main.ts notebooks create --name GARY --json      # → { "id": "<uuid>", "url": "…/notebook/<uuid>" }

# 2) Re-ingest the base context that lives on disk (candidate-agnostic; regenerated by onboarding):
npx -y bun scripts/main.ts sources add-file data/cv/base/<cv-file>.pdf   --notebook <uuid>
npx -y bun scripts/main.ts sources add-file data/cv/base/gary-context.md --notebook <uuid>

Then write the new <uuid> back to GARY's onboarding state so the app opens the right notebook (the desktop app persists it as notebookId; from the app the ingest emits a gary://ingest {kind:"notebook", id} event). Notes:

• Auth reuses the ONE automation browser :9333 (the same one all job-board connections use — it already holds the Google/Gmail session). GARY bridges its cookies via engines/notebooklm-session.mjs → login --cookies "<header>". Do NOT spawn a second browser or run login (which polls for 5 min). Just being logged into Gmail is not enough — the browser must have entered NotebookLM (passed Google's account chooser).
• The RPC auth works because rpc-client.ts init() follows redirects (refreshing the rotating __Secure-1PSIDTS); only the long-lived __Secure-1PSID must stay valid.
• Host (NotebookLM → Gemini migration): the app moved to notebook.google.com. The legacy notebooklm.google.com still serves the app HTML via redirect, but every batchexecute RPC posted there answers 401 — that's why notebooks create blew up while login/status looked healthy. The host is now resolved at runtime (scripts/host.ts): init() pins the origin where the redirect chain lands and all requests (RPC, chat stream, upload, media) target it. Never hardcode a host; if Google moves it again this keeps working. engines/notebooklm-session.mjs validates by PROBING a real read-only RPC (wXbhsf) on the resolved origin — "the page loaded" is not proof the session authorizes RPCs.
• login NUNCA guarda una sesión sin probarla (scripts/session.ts → assertSessionAuthorizes: init()
  • un wXbhsf real). Tener SID + __Secure-1PSID NO prueba nada — sobreviven a una sesión muerta y notebook.google.com/login también es "una URL de NotebookLM". Antes eso hacía que "Verificar sesión" dijera OK, sobrescribiera cookies.json con cookies muertas, y el fallo apareciera tarde en notebooks create → "NotebookLM redirected to Google login". Ahora los tres caminos validan: login --cookies <hdr> (puente :9333) rechaza y sale 1 sin tocar el archivo; login sin --force re-loguea si la sesión guardada ya no autoriza; login --force sigue sondeando hasta que el probe pase. Además se guarda el jar REFRESCADO (Google rota durante el ).

Notebook Management

# List all notebooks in library
npx -y bun scripts/main.ts notebooks list

# Add a notebook by URL
npx -y bun scripts/main.ts notebooks add <url> [--name <name>] [--description <desc>] [--topics <t1,t2>]

# Remove a notebook from library
npx -y bun scripts/main.ts notebooks remove <id>

# Set active notebook (used as default for generation)
npx -y bun scripts/main.ts notebooks activate <id>

# Search notebooks
npx -y bun scripts/main.ts notebooks search <query>

Chat with Notebook AI

# Ask a question about your notebook sources
npx -y bun scripts/main.ts chat --question "What are the key findings?" --notebook abc123

# Continue a conversation
npx -y bun scripts/main.ts chat --question "Tell me more about that" --conversation-id <id>

# Get JSON output with citations
npx -y bun scripts/main.ts chat --question "Summarize the methodology" --json

Source Management

# List all sources in a notebook
npx -y bun scripts/main.ts sources list --notebook abc123

# Add sources
npx -y bun scripts/main.ts sources add-url https://example.com/article --notebook abc123
npx -y bun scripts/main.ts sources add-youtube https://youtube.com/watch?v=xxx --notebook abc123
npx -y bun scripts/main.ts sources add-text --title "My Notes" --content "Important findings..." --notebook abc123
npx -y bun scripts/main.ts sources add-file ./paper.pdf --notebook abc123

# Delete a source
npx -y bun scripts/main.ts sources delete <sourceId> --notebook abc123

Supported file types: PDF, TXT, MD, DOCX, CSV, EPUB, images (PNG, JPG, WEBP, etc.), audio/video files.

Research

# Fast web research (finds relevant sources quickly)
npx -y bun scripts/main.ts research fast --query "latest AI agent frameworks" --notebook abc123

# Deep research (comprehensive report with analysis)
npx -y bun scripts/main.ts research deep --query "state of LLM reasoning" --notebook abc123

# Auto-import found sources into the notebook
npx -y bun scripts/main.ts research fast --query "topic" --notebook abc123 --import

# Check research status
npx -y bun scripts/main.ts research status --notebook abc123

Notes

# List notes in a notebook
npx -y bun scripts/main.ts notes list --notebook abc123

# Create a note
npx -y bun scripts/main.ts notes create --title "Key Takeaways" --content "1. Finding one..." --notebook abc123

# Update a note
npx -y bun scripts/main.ts notes update <noteId> --title "Updated Title" --content "New content" --notebook abc123

# Delete a note
npx -y bun scripts/main.ts notes delete <noteId> --notebook abc123

Artifact Generation

npx -y bun scripts/main.ts generate <type> [options]

Types: slide_deck, audio, video, mind_map, flashcards, quiz, infographic, report, data_table

Options

Global Options

Option	Description
--notebook <url|id>	Notebook URL or library ID (defaults to active notebook)
--output <path>	Output file path (auto-named if omitted)
--instructions <text>	Custom instructions for generation
--json	Output as JSON
--login	Refresh cookies only, then exit
--help, -h	Show help

Slide Deck Options

Option	Description
--format <type>	Output format: pdf (default), pptx

Audio Overview Options

Option	Description
--format <type>	Audio format: deep_dive (default), brief, critique, debate
--length <length>	Duration: short, default, long
--language <lang>	Language code (default: en)

Video Overview Options

Option	Description
--style <style>	Visual style: auto (default), classic, whiteboard, kawaii, anime, watercolor
--format <type>	Video format: explainer (default), brief

Quiz Options

Option	Description
--difficulty <level>	Difficulty: easy, medium (default), hard
--quantity <amount>	Number of questions: fewer, standard, more

Flashcards Options

Option	Description
--difficulty <level>	Difficulty: easy, medium (default), hard
--quantity <amount>	Number of cards: fewer, standard, more

Infographic Options

Option	Description
--orientation <type>	Layout: landscape (default), portrait, square

Report Options

Option	Description
--format <type>	Report format: briefing_doc (default), study_guide, blog_post

Data Table Options

Option	Description
--format <type>	Output format: csv (default), sheets (Google Sheets export)

Artifact Types

Type	Output Format	Description
slide_deck	PDF/PPTX	Presentation slides summarizing notebook sources
audio	M4A	Audio overview in conversation format (deep dive, brief, critique, debate)
video	MP4	Animated video overview with visual styles
mind_map	PNG	Visual mind map of key concepts and relationships
flashcards	HTML/JSON	Study flashcards generated from source material
quiz	HTML/JSON	Multiple-choice quiz with answer key
infographic	PNG	Visual summary infographic in various orientations
report	Markdown	Written report (briefing doc, study guide, blog post)
data_table	CSV/Sheets	Structured data extracted from sources

Authentication

First run opens Chrome to authenticate with Google. Cookies are cached for subsequent runs. Uses CDP browser automation for the login flow, then direct batchexecute RPC calls for all operations.

# Force cookie refresh
npx -y bun scripts/main.ts login --force

Environment variables

Variable	Description
NOTEBOOKLM_DATA_DIR	Data directory override
NOTEBOOKLM_COOKIE_PATH	Cookie file path
NOTEBOOKLM_CHROME_PROFILE_DIR	Chrome profile directory
NOTEBOOKLM_OUTPUT_DIR	Default output directory

Rate Limits

NotebookLM free tier limits:

Resource	Limit
Audio/video overviews	3 per day
Reports/flashcards/quizzes	10 per day
Daily chats	50
Total notebooks	100
Sources per notebook	50

Examples

Generate a slide deck from a specific notebook

npx -y bun scripts/main.ts generate slide_deck \
  --notebook https://notebooklm.google.com/notebook/abc123 \
  --output presentation.pdf

Create an audio deep dive with custom instructions

npx -y bun scripts/main.ts generate audio \
  --format deep_dive \
  --length long \
  --instructions "Focus on the methodology section and compare approaches"

Generate a whiteboard-style video overview

npx -y bun scripts/main.ts generate video \
  --style whiteboard \
  --length medium \
  --output explainer.mp4

Create a quiz and get JSON output for integration

npx -y bun scripts/main.ts generate quiz \
  --difficulty hard \
  --quantity more \
  --json > quiz_data.json

Generate a portrait infographic

npx -y bun scripts/main.ts generate infographic \
  --orientation portrait \
  --instructions "Highlight the three main findings" \
  --output summary.png

Export structured data as CSV

npx -y bun scripts/main.ts generate data_table \
  --format csv \
  --output extracted_data.csv