teng-lin/notebooklm-py

Unofficial Python API and agentic skill for Google Gemini Notebook. Full programmatic access to NotebookLM's features—including capabilities the web UI doesn't expose—via Python, CLI, and AI agents like Claude Code, Codex, and OpenClaw.

What it solves

notebooklm-py provides programmatic access to Google's NotebookLM (now Gemini Notebook), allowing users to bypass the web UI to automate research, content generation, and knowledge management. It enables the integration of NotebookLM's grounded AI capabilities into AI agents (like Claude Code or Codex), shell scripts, and custom Python pipelines, effectively turning the service into a scriptable "memory layer" or synthesis engine for other LLMs.

How it works

The library acts as an unofficial Python API and CLI wrapper around undocumented Google APIs. It supports multiple authentication methods, including interactive browser login, cookie imports from existing browsers, and a master-token system for headless server environments. It can interact with NotebookLM to manage notebooks, upload various source types (URLs, PDFs, YouTube videos, Google Drive files), and trigger the generation of cited answers and multimedia artifacts.

Who it’s for

  • AI Agent Developers: Those wanting to give agents persistent, grounded memory or a "project brain" without managing their own vector databases.
  • Researchers and Students: Users needing to automate the bulk import of sources and the generation of study aids like quizzes, flashcards, and podcasts.
  • Automation Engineers: People building headless pipelines for scheduled audio briefings or automated incident runbook generation.
  • Knowledge Managers: Users syncing NotebookLM's synthesized outputs into tools like Obsidian.

Highlights

  • Full Feature Coverage: Programmatic control over notebooks, sources, chat, notes, and sharing permissions.
  • Diverse Artifact Generation: Generates audio overviews (podcasts), videos, slide decks, infographics, quizzes, flashcards, reports, and data tables.
  • Extended Capabilities: Offers features not found in the web UI, such as batch downloads of artifacts and structured exports (JSON, CSV, Markdown) for quizzes and mind maps.
  • Agent-Ready: Includes built-in skills for Claude Code and Codex, and an MCP server for integration with Claude Desktop and ChatGPT.
  • Flexible Auth: Supports multi-account profiles and a master-token system for unattended, headless operation.

Related

  • Project
    claude-world/notebooklm-skill

    An automation tool and MCP server for Google NotebookLM that enables creating notebooks, performing grounded research, and generating artifacts via CLI or AI agents.

  • Project
    jacob-bd/gemini-notebook-mcp-cli

    A Python package that provides a command‑line tool (`nlm`) and a Model Context Protocol (MCP) server for programmatic access to Google’s Gemini Notebook. It lets you list/create notebooks, add sources, run AI‑driven queries, generate audio/video/slide content, download artifacts, share notebooks, and integrate the service with AI assistants (Claude, Cursor, Gemini‑CLI, etc.) via automatically generated MCP configurations.

  • Dispatch
    NotebookLM Rebranded as Gemini Notebook

    Google has rebranded NotebookLM to Gemini Notebook to align the product with the broader Gemini AI ecosystem, sparking mixed reactions regarding brand consistency and product longevity.

  • Project
    tmc/nlm

    A single-binary Go client for NotebookLM that provides a scriptable CLI, an interactive chat client, and an MCP server for AI agents.

  • Project
    codejunkie99/agentic-stack

    agentic‑stack is an open‑source toolbox that gives a portable “brain” (memory, skills, protocols) to many code‑generation agents (Claude Code, Cursor, Copilot CLI, Gemini CLI, etc.). The brain lives in a hidden `.agent/` folder, enabling agents to retain preferences, lessons, and reusable skills across tool switches. It provides bounded agentic loops, a review‑by‑human protocol, a local dashboard for token‑cost/KPI monitoring, and export utilities that turn approved runs into trace records or training‑ready JSONL. Installable via Homebrew or a simple script, it supports optional external long‑term memory (the `brain` server) and a beta full‑text memory search.