Control NotebookLM from Claude Code

---
name: notebooklm
description: Complete API for Google NotebookLM - full programmatic access including features not in the web UI. Create notebooks, add sources, generate all artifact types, download in multiple formats. Activates on explicit /notebooklm or intent like "create a podcast about X", "install notebooklm", "add notebooklm to cowork"
---
<!-- notebooklm-py v0.3.4 -->

# NotebookLM Automation

Complete programmatic access to Google NotebookLM—including capabilities not exposed in the web UI. Create notebooks, add sources (URLs, YouTube, PDFs, audio, video, images), chat with content, generate all artifact types, and download results in multiple formats.

## Step 0: Setup (Run Automatically on First Use)

When this skill is triggered and `notebooklm` is not yet installed or authenticated, complete setup first.

### Pre-flight: Check Python Version

`notebooklm-py` requires **Python 3.10+**. Check the available version before installing:

```bash
python3 --version
```

If Python is below 3.10 (e.g. 3.9.x which is the macOS default), install a compatible version:

**macOS (Homebrew):**
```bash
brew install python@3.12
```
Then use `/opt/homebrew/bin/python3.12` (Apple Silicon) or `/usr/local/bin/python3.12` (Intel) for the venv below.

**Linux (apt):**
```bash
sudo apt update && sudo apt install -y python3.12 python3.12-venv
```

### Install the CLI

Always use a virtual environment to avoid "externally-managed-environment" errors and PATH issues.

Determine which Python to use — if the system `python3` is 3.10+, use it directly. Otherwise use the one you just installed (e.g. `python3.12`):

```bash
# Set PYTHON to the correct binary (adjust if needed)
PYTHON=$(command -v python3.12 2>/dev/null || command -v python3.11 2>/dev/null || command -v python3.10 2>/dev/null || command -v python3)

# Verify it's 3.10+
$PYTHON -c "import sys; assert sys.version_info >= (3,10), f'Python {sys.version} is too old — need 3.10+'; print(f'Using Python {sys.version}')"

# Create venv and install
$PYTHON -m venv ~/.notebooklm-venv
source ~/.notebooklm-venv/bin/activate
pip install "notebooklm-py[browser]"
playwright install chromium
```

Then symlink so it's always on PATH:
```bash
mkdir -p ~/bin
ln -sf ~/.notebooklm-venv/bin/notebooklm ~/bin/notebooklm
export PATH="$HOME/bin:$PATH"
```

Verify the CLI works:
```bash
notebooklm --help
```

### Authenticate

**IMPORTANT:** The built-in `notebooklm login` command requires interactive terminal input (pressing Enter after sign-in). Claude Code's bash tool does NOT support interactive input, so `notebooklm login` will fail — the browser opens and closes instantly. Instead, use this custom login script.

Tell the user:

> I'm going to open a browser window — just sign into your Google account and navigate to notebooklm.google.com. Take your time, I'll wait for you to confirm before closing it.

Then write and run this login script:

```bash
cat > /tmp/nlm_login.py << 'PYEOF'
import json, os, time
from pathlib import Path
from playwright.sync_api import sync_playwright

STORAGE_PATH = Path.home() / ".notebooklm" / "storage_state.json"
PROFILE_PATH = Path.home() / ".notebooklm" / "browser_profile"
SIGNAL_FILE = Path("/tmp/nlm_save_signal")

SIGNAL_FILE.unlink(missing_ok=True)
STORAGE_PATH.parent.mkdir(parents=True, exist_ok=True)

print("Opening browser for Google login...")
print("Sign in to Google and navigate to notebooklm.google.com")

with sync_playwright() as p:
    browser = p.chromium.launch_persistent_context(
        user_data_dir=str(PROFILE_PATH),
        headless=False,
        args=["--disable-blink-features=AutomationControlled"],
    )
    page = browser.pages[0] if browser.pages else browser.new_page()
    page.goto("https://notebooklm.google.com/")

    print("Browser is open. Waiting for save signal...")
    while not SIGNAL_FILE.exists():
        time.sleep(1)

    print("Save signal received! Capturing session...")
    storage = browser.storage_state()
    with open(STORAGE_PATH, "w") as f:
        json.dump(storage, f)

    cookie_names = [c["name"] for c in storage.get("cookies", [])]
    print(f"Saved {len(cookie_names)} cookies: {cookie_names}")
    browser.close()

SIGNAL_FILE.unlink(missing_ok=True)
print(f"Authentication saved to: {STORAGE_PATH}")
PYEOF

# Run the login script in the background
source ~/.notebooklm-venv/bin/activate
python3 /tmp/nlm_login.py > /tmp/nlm_login_output.txt 2>&1 &
echo "Login started (PID=$!). Browser should open in a few seconds..."
```

Wait ~10 seconds for the browser to open, then ask the user if they can see the browser and are signed in.

Once the user confirms they are on the NotebookLM homepage, save the session:

```bash
touch /tmp/nlm_save_signal
sleep 8
cat /tmp/nlm_login_output.txt
```

Then verify authentication:

```bash
export PATH="$HOME/bin:$PATH"
notebooklm auth check
notebooklm list
```

If auth passes (SID cookie present), confirm to the user that NotebookLM is set up and ready. Clean up the temp script:

```bash
rm -f /tmp/nlm_login.py /tmp/nlm_login_output.txt /tmp/nlm_save_signal
```

If auth fails (SID cookie missing), the user may not have fully signed in. Delete the browser profile and retry:

```bash
rm -rf ~/.notebooklm/browser_profile ~/.notebooklm/storage_state.json
```

Then run the login script again from the top.

---

## Adding NotebookLM to Co-work

When the user asks to "add this to Co-work", "use this in Co-work", or "make this work in Co-work":

### Step 1: Check auth exists

```bash
cat ~/.notebooklm/storage_state.json > /dev/null 2>&1
```

If it doesn't exist, run the install and authenticate steps above first.

### Step 2: Generate the Co-work skill file

Read the contents of this skill file you are currently using. Then read the contents of `~/.notebooklm/storage_state.json`.

**IMPORTANT — Strip cookies before embedding:** Co-work is sandboxed and can't read local files, so cookies must be inlined. But the full storage_state.json contains duplicate cookies across multiple Google domains (.google.ae, .google.co.uk, .youtube.com, etc.) that are NOT needed. Strip them to save ~55% of tokens.

Run this to generate the minimal auth JSON:
```bash
python3 << 'PYEOF'
import json

with open("$HOME/.notebooklm/storage_state.json") as f:
    data = json.load(f)

# Only these domains are needed for NotebookLM auth
essential_domains = {".google.com", "notebooklm.google.com", "accounts.google.com"}
# Skip analytics/tracking cookies
skip_names = {"_gcl_au", "_ga", "_ga_W0LDH41ZCB", "OTZ", "ACCOUNT_CHOOSER"}

stripped = {
    "cookies": [c for c in data["cookies"] if c["domain"] in essential_domains and c["name"] not in skip_names],
    "origins": [o for o in data.get("origins", []) if "notebooklm" in o.get("origin", "")]
}
print(json.dumps(stripped, separators=(',',':')))
PYEOF
```

Create a MODIFIED copy of this skill file where you replace the entire "Step 0: Setup" section and "Adding NotebookLM to Co-work" section with an Auto-Authentication section that inlines the stripped cookies.

### Step 3: Save and instruct

Save the file to the user's Desktop as `NotebookLMSkill-Cowork.md`.

---

## When This Skill Activates

**Explicit:** User says "/notebooklm", "use notebooklm", "install notebooklm", or mentions the tool by name

**Intent detection:** Recognize requests like:
- "Create a podcast about [topic]"
- "Summarise these URLs/documents"
- "Generate a quiz from my research"
- "Turn this into an audio overview"
- "Create flashcards for studying"
- "Generate a video explainer"
- "Make an infographic"
- "Create a mind map of the concepts"

## Autonomy Rules

**Run automatically (no confirmation):**
- `notebooklm status` - check context
- `notebooklm auth check` - diagnose auth issues
- `notebooklm list` - list notebooks
- `notebooklm source list` - list sources
- `notebooklm artifact list` - list artifacts
- `notebooklm language list` - list supported languages
- `notebooklm use <id>` - set context
- `notebooklm create` - create notebook
- `notebooklm ask "..."` - chat queries
- `notebooklm source add` - add sources

**Ask before running:**
- `notebooklm delete` - destructive
- `notebooklm generate *` - long-running, may fail
- `notebooklm download *` - writes to filesystem
- `notebooklm ask "..." --save-as-note` - writes a note

## Quick Reference

| Task | Command |
|------|---------|
| List notebooks | `notebooklm list` |
| Create notebook | `notebooklm create "Title"` |
| Set context | `notebooklm use <notebook_id>` |
| Add URL source | `notebooklm source add "https://..."` |
| Add file | `notebooklm source add ./file.pdf` |
| Add YouTube | `notebooklm source add "https://youtube.com/..."` |
| Web research | `notebooklm source add-research "query"` |
| Chat | `notebooklm ask "question"` |
| Generate podcast | `notebooklm generate audio "instructions"` |
| Generate video | `notebooklm generate video "instructions"` |
| Generate report | `notebooklm generate report --format briefing-doc` |
| Generate quiz | `notebooklm generate quiz` |
| Generate flashcards | `notebooklm generate flashcards` |
| Generate infographic | `notebooklm generate infographic` |
| Generate mind map | `notebooklm generate mind-map` |
| Generate slide deck | `notebooklm generate slide-deck` |
| Download audio | `notebooklm download audio ./output.mp3` |
| Download video | `notebooklm download video ./output.mp4` |
| Download slides | `notebooklm download slide-deck ./slides.pptx --format pptx` |

## Generation Types

| Type | Command | Formats | Download |
|------|---------|---------|----------|
| Podcast | `generate audio` | deep-dive, brief, critique, debate | .mp3 |
| Video | `generate video` | explainer, brief + 9 visual styles | .mp4 |
| Slide Deck | `generate slide-deck` | detailed, presenter | .pdf / .pptx |
| Report | `generate report` | briefing-doc, study-guide, blog-post, custom | .md |
| Quiz | `generate quiz` | easy, medium, hard | .json / .md / .html |
| Flashcards | `generate flashcards` | easy, medium, hard | .json / .md / .html |
| Infographic | `generate infographic` | landscape, portrait, square | .png |
| Mind Map | `generate mind-map` | (instant) | .json |
| Data Table | `generate data-table` | description required | .csv |

## Error Handling

| Error | Cause | Action |
|-------|-------|--------|
| Auth/cookie error | Session expired | Re-authenticate |
| "No notebook context" | Context not set | Run `notebooklm use <id>` |
| Rate limiting | Google throttle | Wait 5-10 min, retry |
| Download fails | Generation incomplete | Check `artifact list` for status |

## Known Limitations

- Audio, video, quiz, flashcard, infographic, and slide deck generation may fail due to Google rate limits
- Generation times: audio 10-20 min, video 15-45 min, quiz/flashcards 5-15 min
- This is an unofficial API — Google can change things without warning