From notebooklm
Provides NotebookLM integration patterns for notebooklm-mcp: auth setup, listing/creating/querying notebooks, managing sources, research workflows, and studio artifact generation (audio/video/slides/infographics).
npx claudepluginhub ray-manaloto/notebooklm-claude-integration --plugin notebooklmThis skill uses the workspace's default tool permissions.
Master NotebookLM integration for source-grounded research with citation-backed answers from Gemini.
Queries Google NotebookLM notebooks for source-grounded, citation-backed Gemini answers. Manages auth, adds/lists notebooks via Python scripts with browser automation.
Queries Google NotebookLM notebooks for source-grounded, citation-backed Gemini answers using browser automation, library management, and persistent auth.
Integrates Google NotebookLM via nlm CLI for querying project docs, managing notebooks/sources, retrieving AI-synthesized info, and generating podcasts/reports. Use for RAG on curated knowledge bases.
Share bugs, ideas, or general feedback.
Master NotebookLM integration for source-grounded research with citation-backed answers from Gemini.
notebook_queryAsk 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:
notebook_list: list all notebooksnotebook_create: create a new notebooknotebook_get: get notebook details with sourcesnotebook_describe: AI summary of notebook contentnotebook_rename: rename a notebooknotebook_delete: delete a notebook (confirmation required)notebook_add_url: add URL/YouTube as sourcenotebook_add_text: add pasted text as sourcenotebook_add_drive: add Google Drive doc as sourcesource_list_drive: list sources with freshness statussource_sync_drive: sync stale Drive sources (confirmation required)source_delete: delete a source (confirmation required)source_describe: AI summary and keywords for a sourcesave_auth_tokensPersist 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_start: start web/drive researchresearch_status: poll progress (built-in wait)research_import: import discovered sources into a notebookaudio_overview_createvideo_overview_createinfographic_createslide_deck_createstudio_statusstudio_delete1. Authenticate: run `notebooklm-mcp-auth`, then `save_auth_tokens`
2. List notebooks: `notebook_list`
3. Query: `notebook_query(notebook_id, question)`
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`
1. `notebook_list` to see all available
2. Select IDs and run `notebook_query` per notebook
3. Synthesize answers with notebook names/IDs labeled
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)
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
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:
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
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
| 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 |