NotebookLM Antigravity Skill
This skill provides seamless integration with Google NotebookLM through the nlm command-line interface. It enables you to create notebooks, add sources from URLs, text, files, or Google Drive, generate AI-powered audio podcasts, perform web research, and query your notebooks with natural language.
Goal
To enable Google NotebookLM functionality within Antigravity for research, content generation, and knowledge management tasks.
Prerequisites
- Install the tool: Run the installation script before first use
- Authenticate: You must authenticate with your Google account before using NotebookLM features
Instructions
Step 1: Installation (First Time Only)
Before using NotebookLM, run the installation script to set up the nlm CLI:
bash scripts/install.sh
This will install notebooklm-mcp-cli using uv (recommended) or pip as a fallback.
After installation, verify it's available:
nlm --version
Step 2: Authentication
You must authenticate with your Google account before using NotebookLM:
Auto Mode (Recommended):
nlm login
This launches a dedicated Chrome profile. Log in to Google, and cookies will be extracted automatically.
Check Authentication Status:
nlm login --check
Use Multiple Profiles:
nlm login --profile work
nlm login --profile personal
Step 3: Common Operations
Once authenticated, you can use any of these commands directly in the terminal:
Notebook Management
# List all notebooks
nlm notebook list
# Create a new notebook
nlm notebook create "Research Project"
# Get notebook details
nlm notebook get <notebook-id>
Add Sources
# Add URL source
nlm source add <notebook-id> --url "https://example.com/article"
# Add YouTube video
nlm source add <notebook-id> --url "https://youtube.com/watch?v=xxx"
# Add text directly
nlm source add <notebook-id> --text "Your text content here"
# Add local file
nlm source add <notebook-id> --file /path/to/file.pdf
# Import from Google Drive
nlm source add <notebook-id> --drive "document-id"
Query Notebooks (AI Chat)
# Ask questions about your notebook
nlm notebook query <notebook-id> "What are the key findings?"
# Generate summary
nlm notebook query <notebook-id> "Summarize all sources"
Generate Audio Podcasts
# Generate audio overview
nlm audio create <notebook-id> --confirm
# Download generated audio
nlm download audio <notebook-id> <artifact-id>
# Download as MP3
nlm download audio <notebook-id> <artifact-id> --format mp3
Generate Other Studio Content
# Generate video explainer
nlm studio create <notebook-id> --type video --style classic
# Generate briefing document
nlm studio create <notebook-id> --type briefing
# Generate flashcards
nlm studio create <notebook-id> --type flashcard --difficulty medium
# Generate infographic
nlm studio create <notebook-id> --type infographic --orientation landscape
# Generate mind map
nlm studio create <notebook-id> --type mindmap
# Generate slide deck
nlm studio create <notebook-id> --type slideshow
Web Research
# Start web research on a topic
nlm research start "enterprise AI ROI metrics"
# Deep research with auto-import
nlm research start "cloud marketplace trends" --deep --import
Share Notebooks
# Make notebook public
nlm share public <notebook-id>
# Disable public access
nlm share disable <notebook-id>
# Invite collaborators
nlm share invite <notebook-id> --email user@example.com --role editor
Sync Google Drive Sources
# Sync all Drive sources
nlm source sync drive <notebook-id>
# Check which sources need sync
nlm source list <notebook-id>
Step 4: Check Studio Content Status
After generating studio content (audio, video, etc.), check the status:
# List all studio artifacts
nlm studio list <notebook-id>
# Get status of specific artifact
nlm studio status <notebook-id> <artifact-id>
Poll the status until it shows "completed", then download using the download command.
Examples
Example 1: Research and Generate Podcast
# Create notebook for research
NOTEBOOK_ID=$(nlm notebook create "AI Research 2026" | grep -oP 'id: \K\S+')
# Add several sources
nlm source add "$NOTEBOOK_ID" --url "https://arxiv.org/abs/2301.07041"
nlm source add "$NOTEBOOK_ID" --url "https://example.com/ai-report"
# Generate audio podcast
nlm audio create "$NOTEBOOK_ID" --confirm
# Wait and download
nlm download audio "$NOTEBOOK_ID" <artifact-id>
Example 2: Quick URL Analysis
# Create notebook, add URL, and query in one flow
NOTEBOOK_ID=$(nlm notebook create "Quick Analysis" | grep -oP 'id: \K\S+')
nlm source add "$NOTEBOOK_ID" --url "https://example.com/article"
nlm notebook query "$NOTEBOOK_ID" "Summarize the main arguments in 3 bullet points"
Example 3: File-Based Research
# Add local PDF and query
nlm notebook create "Document Review"
nlm source add <notebook-id> --file /path/to/report.pdf
nlm notebook query <notebook-id> "What are the key recommendations?"
Example 4: Web Research with Auto-Import
# Research and automatically import top sources
nlm research start "AI agents productivity metrics" --deep --import
# This will create a notebook and import the best sources found
Constraints
IMPORTANT:
- Authentication is required before any NotebookLM operations
- Use
nlm login --checkto verify authentication status - Free tier accounts have rate limits (~50 queries/day)
- Cookie-based auth expires every 2-4 weeks; re-run
nlm loginwhen prompted - Always check studio content status before downloading (generation takes time)
- Be mindful of context limits when querying large notebooks
Safety:
- This tool uses undocumented internal APIs (as documented by notebooklm-mcp-cli)
- Use for personal/experimental purposes only
- Cookie extraction is handled securely by the
nlm logincommand
Troubleshooting
Installation Issues:
# Force reinstall if needed
uv tool install --force notebooklm-mcp-cli
Authentication Failures:
# Remove old auth data and re-login
rm -rf ~/.notebooklm-mcp-cli
nlm login
Command Not Found:
# Verify installation
which nlm
# If not found, run install.sh again
Token Expiration:
# Refresh auth
nlm login
Advanced Usage
Profile Management:
# List all profiles with emails
nlm login profile list
# Switch profiles
nlm login switch work
# Delete a profile
nlm login profile delete work
Using uvx (No Install Required):
# Run nlm without installing
uvx --from notebooklm-mcp-cli nlm notebook list
References
For complete command reference and advanced features, see:
- CLI Guide: Available in notebooklm-mcp-cli package or GitHub
- MCP Guide: https://github.com/jacob-bd/notebooklm-mcp-cli