notebooklm-patterns

name	notebooklm-patterns
description	NotebookLM integration patterns aligned to notebooklm-mcp (RPC tool names, auth flow, and troubleshooting). Use when listing/creating/querying notebooks, managing sources, running research, or generating studio artifacts.

NotebookLM Integration Patterns

Master NotebookLM integration for source-grounded research with citation-backed answers from Gemini.

When to Use This Skill

• Setting up NotebookLM authentication for notebooklm-mcp
• Listing/creating/querying notebooks and managing sources
• Running research workflows (web/drive) and importing sources
• Generating studio artifacts (audio, video, slides, infographics)
• Troubleshooting authentication, timeouts, or rate limits

MCP Tools Reference

Query Tools (RPC)

notebook_query

Ask a question to a specific notebook by notebook_id.

Parameters:
- notebook_id (required): Notebook ID from notebook_list
- question (required): The question to ask
- session_id (optional): For contextual follow-up questions

Usage Pattern:

# Query specific notebook by ID
notebook_query(
    notebook_id="abc123",
    question="What topics are covered?"
)

# Continue a session
notebook_query(
    notebook_id="abc123",
    question="Show an example.",
    session_id="previous_session_id"
)

Timeout Handling:

• If a notebook query times out, retry once.
• If it still times out, record a timeout for that notebook and continue.

Library & Source Tools (RPC)

Notebooks

• notebook_list: list all notebooks
• notebook_create: create a new notebook
• notebook_get: get notebook details with sources
• notebook_describe: AI summary of notebook content
• notebook_rename: rename a notebook
• notebook_delete: delete a notebook (confirmation required)

Sources

• notebook_add_url: add URL/YouTube as source
• notebook_add_text: add pasted text as source
• notebook_add_drive: add Google Drive doc as source
• source_list_drive: list sources with freshness status
• source_sync_drive: sync stale Drive sources (confirmation required)
• source_delete: delete a source (confirmation required)
• source_describe: AI summary and keywords for a source

Authentication Tools (RPC)

save_auth_tokens

Persist auth cookies for notebooklm-mcp. Required before using RPC tools.

Default flow: run notebooklm-mcp-auth once, then save_auth_tokens.

File mode: use notebooklm-mcp-auth --file to manually extract cookies, then save_auth_tokens.

Research & Studio Tools (RPC)

Research

• research_start: start web/drive research
• research_status: poll progress (built-in wait)
• research_import: import discovered sources into a notebook

Studio

• audio_overview_create
• video_overview_create
• infographic_create
• slide_deck_create
• studio_status
• studio_delete

Workflow Patterns

Pattern 1: First-Time Setup

1. Authenticate: run `notebooklm-mcp-auth`, then `save_auth_tokens`
2. List notebooks: `notebook_list`
3. Query: `notebook_query(notebook_id, question)`

Pattern 2: Smart Notebook Discovery

Before adding a notebook, discover its content:

1. Query by notebook ID with `notebook_query`
2. Use response to name/describe with `notebook_rename` or `notebook_describe`

Pattern 3: Multi-Notebook Research

1. `notebook_list` to see all available
2. Select IDs and run `notebook_query` per notebook
3. Synthesize answers with notebook names/IDs labeled

Pattern 4: Session Continuity

For follow-up questions in the same context:

1. First query: response = notebook_query(question)
2. Extract session_id from response
3. Follow-up: notebook_query(follow_up, session_id=session_id)

Troubleshooting Guide

Authentication Issues

Problem: "Not authenticated" error

Solution:
1. Run notebooklm-mcp-auth (auto mode) and log in once
2. Call save_auth_tokens
3. Retry the RPC call

Problem: Authentication keeps failing

Solution:
1. Re-run notebooklm-mcp-auth (auto or file mode)
2. Call save_auth_tokens

Problem: Session expired

Solution:
1. Re-run notebooklm-mcp-auth
2. Call save_auth_tokens

Problem: CDP not detecting Chrome

Solution:
1. Ensure Chrome started with --remote-debugging-port=9222
2. Check if port is listening: curl http://localhost:9222/json/version
3. Only one Chrome instance can use debugging port at a time
4. Close other Chrome instances and restart with flag

Rate Limiting

Problem: "Rate limit reached" (50 queries/day free tier)

Solutions:
1. Wait until daily reset
2. Re-run notebooklm-mcp-auth with a different Google account
3. Upgrade to Google AI Pro/Ultra for 5x limits

Rate Limit Details:

• Free tier: 50 queries/day, 100 notebooks, 50 sources each
• Pro/Ultra: 250 queries/day, 500 notebooks

Browser Issues

Problem: Chrome crashes or hangs

Solution:
1. Close all Chrome instances
2. Clear browser state:
   cleanup_data(confirm=true, preserve_library=true)
3. Re-run notebooklm-mcp-auth and call save_auth_tokens

Problem: "Browser not installed" error

Solution:
1. Ensure Chrome (not Chromium) is installed
2. Check browser_install tool if available

Query Issues

Problem: Empty or poor responses

Solutions:
1. Rephrase question to be more specific
2. Check notebook has relevant content
3. Try different notebook from library

Problem: Wrong notebook being queried

Solution:
1. Use notebook_list to see notebooks
2. Re-run query with the correct notebook_id

Best Practices

Notebook Organization

1. Descriptive Names: Use clear, searchable names
2. Specific Topics: Tag with precise topics for search
3. One Domain Per Notebook: Don't mix unrelated content
4. Regular Cleanup: Remove outdated notebooks

Query Optimization

1. Specific Questions: "How do I implement X?" not "Tell me about X"
2. Follow-up Sessions: Use session_id for related questions
3. Multiple Angles: Ask same topic from different perspectives
4. Verify Sources: Check citations in responses

Research Workflows

1. Discovery First: Always discover content before adding
2. Organize by Project: Group notebooks by use case
3. Regular Updates: Re-discover notebooks when source content changes
4. Cross-Reference: Query multiple notebooks for comprehensive answers

Limits and Quotas

Resource	Free Tier	Pro/Ultra
Daily Queries	50	250
Notebooks	100	500
Sources per Notebook	50	100
Words per Source	500,000	1,000,000

Security Notes

• All credentials stored locally
• Browser state encrypted
• Never share browser_state directory
• Consider dedicated Google account for automation
• NotebookLM terms of service apply