Capafy logo Capafy
Inizia
notebooklm

notebooklm

by: cheahhl814
Programmatic access to Google NotebookLM via the nlm CLI — manages notebooks, sources, queries, and artifacts with citation-quality source titles.
#Ricerca#Istruzione#Produttività
Valutazione
Servono altre valutazioni
Venduti
0
Come si usa
Scarica

notebooklm

Programmatic access to Google NotebookLM via the nlm CLI — manages notebooks, sources, queries, and artifacts with citation-quality source titles.

Features

  • Notebook management — create, list, describe notebooks
  • Source management — add URLs, text, Drive docs; delete broken sources
  • One-shot queries — nlm notebook query <id> "question"
  • Multi-step research — nlm research start discovers new sources
  • Artifact generation — audio, reports, quizzes, flashcards, mindmaps, slides
  • Citation forcing — SYSTEM prompt prevents untraceable [1], [2] citations

Install & Authenticate

pip install notebooklm-mcp
nlm auth login     # opens browser for Google OAuth
nlm auth status    # verify session
nlm doctor         # debug issues

Quick Start

# List notebooks
nlm notebook list --json

# Add sources with citation-quality titles
nlm source add <id> --url "https://doi.org/10.xxxx/xxxxx" --title "Smith et al. 2024 - Paper Title"

# Query with forced source-title citations
nlm notebook query <id> "What gaps exist in mangrove fisheries research?"

# Generate a report artifact
nlm report create <id> --confirm

Citation Quality System

NotebookLM defaults to numbered citations [1], [2] that are untraceable. This skill forces source-title citations by prepending a SYSTEM instruction:

SYSTEM: Do NOT use [1], [2] citations. They are untraceable.
Instead, after EVERY factual claim, write the FULL SOURCE TITLE in parentheses.

Always add sources with "Author Year - Paper Title" format:

# ✅ Good
nlm source add <id> --url "..." --title "Smith et al. 2024 - Mangrove fisheries decline"

# ❌ Bad (raw URL as title)
nlm source add <id> --url "..." --title "https://link.springer.com/article/10.xxxx"

# ❌ Bad (CAPTCHA title)
nlm source add <id> --url "..." --title "Checking your browser - reCAPTCHA"

Key Commands

Command Purpose
nlm notebook list --json List all notebooks with UUIDs
nlm notebook describe <id> AI-generated notebook summary
nlm source add <id> --url "..." Add web/YouTube source
nlm source add <id> --text "..." --title "..." Add pasted text
nlm source add <id> --drive <doc-id> Add Google Drive doc
nlm source content <id> Retrieve raw indexed text
nlm source delete "<id>" --confirm Remove broken source
nlm notebook query <id> "question" One-shot grounded answer
nlm research start "query" --notebook-id <id> Multi-step research
nlm <type> create <id> --confirm Generate artifact (audio, report, quiz, etc.)
nlm alias set <name> <id> Create notebook alias

Artifact Types

audio, report, quiz, flashcards, mindmap, slides, infographic, video, data-table

⚠️ Artifacts can only be created via CLI. Download/export requires the NotebookLM web interface.

Known Limitations

  • Sessions expire after ~20 minutes — re-run nlm login on auth errors
  • Free tier: ~50 queries/day
  • No CLI download/export of artifacts — web UI only
  • Raw URLs and CAPTCHA pages pollute source titles — always use "Author Year - Title" format

Requirements

  • Python 3.10+
  • notebooklm-mcp package (pip install notebooklm-mcp)
  • Google account with NotebookLM access
  • nlm auth login completed before use

View more