NotebookLM Research & Content Generation

When to Use This Skill

•User asks for research on fraud, scams, cybersecurity topics
•User wants to generate content (reports, podcasts, flashcards, quizzes)
•User mentions NotebookLM, notebooks, or sources
•User needs to analyze documents or URLs for QadamSafe content
•User wants to create educational materials about fraud prevention

QadamSafe Context

This skill supports the QadamSafe project — a fraud prevention platform for Kazakhstan. Common use cases:

•Researching new fraud schemes for scenario creation
•Generating educational content about moshennicherstvo (мошенничество)
•Creating audio/video overviews for elderly users
•Building quizzes and flashcards for training

Available Notebooks

Check existing notebooks before creating new ones:

code

mcp_notebooklm_notebook_list(max_results=20)

Known QadamSafe notebook: ce114da7-885a-470d-9431-cdcd5cb5a67c (Мошенничество в Казахстане)

Core Workflows

1. Research New Topics (Web Search)

Use when user asks: "исследуй...", "найди информацию о...", "research..."

python

# Step 1: Start research
mcp_notebooklm_research_start(
    query="телефонное мошенничество Казахстан 2026",
    source="web",  # or "drive"
    mode="fast",   # "fast" (~30s, ~10 sources) or "deep" (~5min, ~40 sources)
    notebook_id="ce114da7-885a-470d-9431-cdcd5cb5a67c"  # Optional: add to existing
)

# Step 2: Poll for completion
mcp_notebooklm_research_status(
    notebook_id="<notebook_id>",
    max_wait=300
)

# Step 3: Import sources (after status="completed")
mcp_notebooklm_research_import(
    notebook_id="<notebook_id>",
    task_id="<task_id>"
)

2. Generate Reports

Use when user asks: "создай отчёт", "сгенерируй документ", "briefing..."

python

# Available formats: "Briefing Doc", "Study Guide", "Blog Post", "Create Your Own"
mcp_notebooklm_report_create(
    notebook_id="ce114da7-885a-470d-9431-cdcd5cb5a67c",
    report_format="Briefing Doc",
    language="ru",  # ru, en, kk
    confirm=True    # REQUIRED for execution
)

3. Generate Audio Overview (Podcast)

Use when user asks: "создай подкаст", "аудио обзор", "podcast..."

python

mcp_notebooklm_audio_overview_create(
    notebook_id="<notebook_id>",
    format="deep_dive",  # deep_dive|brief|critique|debate
    length="default",    # short|default|long
    language="ru",
    focus_prompt="Объясни простым языком для пожилых людей",
    confirm=True
)

4. Generate Educational Content

Flashcards:

python

mcp_notebooklm_flashcards_create(
    notebook_id="<notebook_id>",
    difficulty="medium",  # easy|medium|hard
    confirm=True
)

Quiz:

python

mcp_notebooklm_quiz_create(
    notebook_id="<notebook_id>",
    question_count=10,
    difficulty="medium",
    confirm=True
)

Mind Map:

python

mcp_notebooklm_mind_map_create(
    notebook_id="<notebook_id>",
    title="Виды мошенничества",
    confirm=True
)

5. Add Sources to Notebook

From URL (website/YouTube):

python

mcp_notebooklm_notebook_add_url(
    notebook_id="<notebook_id>",
    url="https://example.com/article"
)

From Text:

python

mcp_notebooklm_notebook_add_text(
    notebook_id="<notebook_id>",
    text="Содержимое текста...",
    title="Название источника"
)

From Google Drive:

python

mcp_notebooklm_notebook_add_drive(
    notebook_id="<notebook_id>",
    document_id="1a2b3c4d5e",  # From Drive URL
    title="Document Title",
    doc_type="doc"  # doc|slides|sheets|pdf
)

6. Query Notebook Content

Use when user asks questions about existing sources:

python

mcp_notebooklm_notebook_query(
    notebook_id="<notebook_id>",
    query="Какие основные виды телефонного мошенничества?"
)

7. Check Generation Status

After creating audio/video, check status:

python

mcp_notebooklm_studio_status(notebook_id="<notebook_id>")

Important Rules

•
Always confirm destructive actions — confirm=True required for:
- •Creating content (reports, audio, flashcards, etc.)
- •Deleting notebooks or sources
- •Syncing Drive sources
•
Language codes — Use BCP-47:
- •Russian: ru
- •English: en
- •Kazakh: kk (may not be fully supported)
•
Research workflow — Always follow: research_start → research_status → research_import
•
Notebook IDs — Always get the notebook ID first with notebook_list or notebook_get
•
Authentication issues — If auth fails, tell user to run:
bash
```
notebooklm-mcp-auth
```

Error Handling

Error	Solution
`401 Unauthorized`	Run `notebooklm-mcp-auth` in terminal
`Notebook not found`	Check notebook ID with `notebook_list`
`Research still running`	Wait and poll `research_status` again
`Source limit reached`	NotebookLM has ~50 source limit per notebook

QadamSafe Integration Ideas

•New Scenario Research: Research recent fraud schemes → Generate report → Extract key points for new scenarios
•Audio for Elderly: Create audio overviews with focus_prompt="Объясни простым для пожилых"
•Training Quizzes: Generate quizzes from fraud materials for gamification
•Weekly Briefings: Auto-generate briefing docs on new fraud trends