notebooklm

NotebookLM Skill

Create notebooks, add sources, analyze content, and generate deliverables via Google NotebookLM.

Prerequisites

• Virtual environment must be set up: bash setup.sh (from project root)
• User must have authenticated: notebooklm login (run in a separate terminal)

Always activate the venv first

cd /home/user/Sneha && source .venv/bin/activate

Available Commands

1. Create a Notebook

python .claude/skills/notebooklm/scripts/create_notebook.py --title "Notebook Title"

Returns JSON: {"notebook_id": "...", "title": "..."}. Save the notebook_id for all subsequent commands.

2. Add YouTube Sources (batch)

python .claude/skills/notebooklm/scripts/add_sources.py \
  --notebook-id <ID> \
  --urls https://youtube.com/watch?v=abc https://youtube.com/watch?v=def ...

Accepts multiple URLs in one call. Adds a 1.5s delay between each to avoid rate limits. Returns summary of added/failed counts.

3. Ask a Question / Analyze

python .claude/skills/notebooklm/scripts/ask_question.py \
  --notebook-id <ID> \
  --question "Summarize the key themes and insights across all sources"

Returns JSON with the analysis answer.

4. Generate an Artifact

python .claude/skills/notebooklm/scripts/generate_artifact.py \
  --notebook-id <ID> \
  --type <ARTIFACT_TYPE> \
  --style "optional style description"

Supported artifact types:

• infographics - Visual infographic
• slide-deck - Presentation slides
• flashcards - Study flashcards
• quiz - Quiz questions
• report - Written report
• mind-map - Mind map diagram
• audio - Audio overview (podcast-style)
• video - Video summary
• data-table - Structured data table

Style examples:

• "handwritten / chalkboard" - Chalkboard aesthetic
• "modern and minimalist" - Clean design
• "colorful and engaging" - Vibrant visuals

End-to-End Workflow (when chaining with yt-research)

1. Receive video list JSON from /yt-research
2. Extract all url values from the results
3. Create a notebook: create_notebook.py --title "[Topic] Research"
4. Add all URLs: add_sources.py --notebook-id <ID> --urls <all URLs>
5. Wait a moment for source processing, then analyze: ask_question.py --notebook-id <ID> --question "..."
6. Generate requested artifact: generate_artifact.py --notebook-id <ID> --type infographics --style "..."
7. Present the analysis and artifact info to the user

Error Handling

If you see an auth error like "session expired" or "not authenticated", tell the user to run notebooklm login again in a separate terminal to re-authenticate.