/query

Query NotebookLM

Ask a question against notebook sources and get an AI-generated answer with citations.

Arguments

Parse from: $ARGUMENTS

Required:

• Question text (the main argument)

Options:

• --notebook <id>: Target a specific notebook (otherwise uses active notebook)
• --source <id>: Limit query to specific source(s) — can be repeated
• --follow-up or -c <conversation_id>: Continue an existing conversation
• --save: Save the answer as a note in the notebook

Workflow

1. Parse arguments: Extract the question, optional notebook ID, source IDs, and conversation ID.

2. Verify context (if no --notebook specified):

   uv run --project "${CLAUDE_PLUGIN_ROOT}/scripts" notebooklm status
   

   Ensure there is an active notebook. If not, suggest running /notebooklm-enhanced:library first.

3. Run the query:

   # Basic query
   uv run --project "${CLAUDE_PLUGIN_ROOT}/scripts" notebooklm ask "YOUR QUESTION HERE" --json
   
   # With specific sources
   uv run --project "${CLAUDE_PLUGIN_ROOT}/scripts" notebooklm ask "YOUR QUESTION" -s SOURCE_ID_1 -s SOURCE_ID_2 --json
   
   # Follow-up in existing conversation
   uv run --project "${CLAUDE_PLUGIN_ROOT}/scripts" notebooklm ask "FOLLOW UP QUESTION" -c CONVERSATION_ID --json
   
   # With explicit notebook
   uv run --project "${CLAUDE_PLUGIN_ROOT}/scripts" notebooklm ask "YOUR QUESTION" --notebook NOTEBOOK_ID --json
   

4. Parse JSON response: The output contains:

   {
     "answer": "The answer text with [1] [2] citation markers...",
     "conversation_id": "...",
     "turn_number": 1,
     "is_follow_up": false,
     "references": [
       {"source_id": "...", "citation_number": 1, "cited_text": "..."},
       {"source_id": "...", "citation_number": 2, "cited_text": "..."}
     ]
   }
   

5. Format output: Present the answer with properly formatted citations. For each reference, show the citation number and the relevant quoted text.

6. Report conversation ID: Always include the conversation_id so the user can run follow-up queries with -c.

Example Usage

/notebooklm-enhanced:query What are the main findings of this paper?
/notebooklm-enhanced:query --notebook abc123 What methods were used?
/notebooklm-enhanced:query --source def456 Summarize this source
/notebooklm-enhanced:query -c conv789 Can you elaborate on point 3?

Notes

• Use --json flag on all ask commands to get structured output with citations.
• The conversation_id enables multi-turn conversations — always surface it to the user.
• If no sources exist in the notebook, the query will fail. Suggest adding sources first.