NotebookLM AI Plugin

Bring the full Google NotebookLM experience to Claude Code, OpenClaw and any other AI agents that execute shell commands. Chat with your notebook AI, generate 9 types of artifacts, manage sources (URLs, YouTube, files), run fast/deep web research, and manage notes — just ask Claude.

Quick Start

Installation

Option 1: CLI Install (Recommended)

Use npx skills to install skills directly:

# Install the skill
npx skills add proyecto26/notebooklm-ai-plugin

# List available skills
npx skills add proyecto26/notebooklm-ai-plugin --list

This automatically installs to your .claude/skills/ directory.

Option 2: Claude Code Plugin

Install via Claude Code's built-in plugin system:

# Add the marketplace
/plugin marketplace add proyecto26/notebooklm-ai-plugin

# Install the plugin
/plugin install notebooklm-ai-plugin

Option 3: Clone and Copy

Clone the repo and copy the skills folder:

git clone https://github.com/proyecto26/notebooklm-ai-plugin.git
cp -r notebooklm-ai-plugin/skills/* .claude/skills/

Option 4: Git Submodule

Add as a submodule for easy updates:

git submodule add https://github.com/proyecto26/notebooklm-ai-plugin.git .claude/notebooklm-ai-plugin

Then reference skills from .claude/notebooklm-ai-plugin/skills/.

Option 5: Fork and Customize

1. Fork this repository
2. Customize the skill for your specific needs
3. Clone your fork into your projects

Authentication

The first time you use the skill, it opens Chrome for Google login. Cookies are cached for subsequent runs. Just tell Claude:

"Log me into NotebookLM"

Generate Your First Artifact

Once authenticated, just describe what you want:

"Add this notebook https://notebooklm.google.com/notebook/YOUR_ID and generate a study guide from it"

That's it — the skill handles everything: authentication, notebook management, artifact creation, polling, and download.

Features

Chat with Notebook AI

Ask questions and get source-grounded answers with citations — the same AI chat experience from NotebookLM's web UI, now in your terminal.

9 Artifact Types

Source Management

Add and manage notebook sources directly from Claude Code — no need to switch to the browser.

• URLs / Websites — Add any web page as a source
• YouTube — Add video transcripts as sources
• File Upload — PDF, TXT, MD, DOCX, CSV, EPUB, images, audio, video
• Pasted Text — Add text content with a custom title
• List / Delete — View and manage existing sources

Fast & Deep Research

Run web research that feeds directly into your notebook:

• Fast Research — Quick web search, finds relevant sources in seconds
• Deep Research — Comprehensive analysis with a full markdown report
• Auto-Import — Automatically add discovered sources to your notebook

Notes Management

Create, update, and delete notes within your notebooks programmatically.

Usage Examples

Just describe what you need to Claude — the skill triggers automatically:

"What are the main findings in my NotebookLM notebook?"

Chats with the notebook AI and returns source-grounded answers with citations.

"Add this YouTube video to my notebook: https://youtube.com/watch?v=..."

Adds the video transcript as a source to your active notebook.

"Upload my research paper paper.pdf to the notebook and generate a study guide"

Uploads the file, then generates a markdown report in study guide format.

"Run a deep research on 'AI agent frameworks 2026' and import the results"

Starts deep web research, waits for completion, and imports sources into the notebook.

"Generate a slide deck from my NotebookLM notebook about machine learning"

Creates a PDF/PPTX presentation from your notebook sources.

"Create a deep dive audio overview of my research papers"

Generates a long-form podcast-style M4A audio discussion.

"Make a portrait infographic highlighting the key findings"

Produces a PNG infographic in portrait orientation.

"Create a note called 'Key Takeaways' with a summary of the main points"

Creates a new note in the notebook with the specified content.

Configuration

Environment Variables

Rate Limits

NotebookLM free tier limits:

Project Structure

notebooklm-ai-plugin/
├── .claude-plugin/
│   ├── plugin.json                  # Plugin manifest
│   └── marketplace.json             # Marketplace metadata
├── skills/
│   └── notebooklm/
│       ├── SKILL.md                 # Skill definition (triggers + docs)
│       └── scripts/
│           ├── main.ts              # CLI entry point (all commands)
│           ├── auth.ts              # Chrome CDP authentication
│           ├── chat.ts              # Streaming chat with notebook AI
│           ├── rpc-client.ts        # batchexecute protocol client
│           ├── rpc-types.ts         # RPC method IDs + enum codes
│           ├── artifact-generator.ts # Create / poll / download artifacts
│           ├── source-manager.ts    # Add / list / delete sources
│           ├── research-manager.ts  # Fast / deep web research
│           ├── notes-manager.ts     # Notes CRUD
│           ├── notebook-manager.ts  # Notebook library CRUD
│           ├── cookie-store.ts      # Cookie persistence
│           ├── paths.ts             # Platform-aware storage
│           ├── types.ts             # Shared TypeScript types
│           └── get-cookie.ts        # Quick login helper
├── LICENSE
└── README.md

For full CLI commands and script usage details, see skills/notebooklm/SKILL.md.

Architecture

graph TB
    subgraph "Claude Code"
        A["User Prompt"] --> B["NotebookLM Skill"]
    end

    subgraph "Authentication"
        B --> C["Chrome CDP"]
        C --> D["Cookie Extraction"]
        D --> E["Cookie Store"]
    end

    subgraph "NotebookLM API"
        E --> F["batchexecute RPC"]
        E --> F2["Streaming Chat"]
        F --> G["Artifacts"]
        F --> H["Sources"]
        F --> I["Research"]
        F --> I2["Notes"]
        F --> I3["Notebooks"]
    end

    subgraph "Capabilities"
        F2 --> S["AI Chat with Citations"]
        G --> J["Slides / Audio / Video"]
        G --> K["Quiz / Flashcards"]
        G --> L["Infographic / Report"]
        H --> M["URL / YouTube / File / Text"]
        I --> N["Fast & Deep Web Research"]
    end

    style A fill:#7c3aed,color:#fff
    style F fill:#1a73e8,color:#fff
    style F2 fill:#1a73e8,color:#fff
    style S fill:#34a853,color:#fff
    style J fill:#34a853,color:#fff
    style K fill:#34a853,color:#fff
    style L fill:#34a853,color:#fff
    style M fill:#ea4335,color:#fff
    style N fill:#fbbc04,color:#000

Hybrid approach: Chrome DevTools Protocol (CDP) handles Google authentication once, then all operations use direct batchexecute RPC calls — no browser overhead for artifact generation.

How It Works

sequenceDiagram
    participant U as User
    participant C as Claude Code
    participant P as Plugin (Bun/TS)
    participant Ch as Chrome (CDP)
    participant N as NotebookLM API

    U->>C: "Generate a slide deck from my notebook"
    C->>P: Invoke notebooklm skill

    Note over P,Ch: First-time auth only
    P->>Ch: Launch Chrome via CDP
    Ch-->>P: Google session cookies
    P->>P: Cache cookies to disk

    P->>N: LIST_NOTEBOOKS (batchexecute RPC)
    N-->>P: Notebook sources
    P->>N: CREATE_ARTIFACT (type=slide_deck)
    N-->>P: Artifact ID

    loop Poll every 5s
        P->>N: LIST_ARTIFACTS
        N-->>P: Status: processing / completed
    end

    P->>N: Download artifact
    N-->>P: PDF/PNG/M4A/MP4 file
    P-->>C: File saved to disk
    C-->>U: "Slide deck saved to slides.pdf"

batchexecute RPC Protocol

The plugin communicates with NotebookLM via Google's internal batchexecute protocol — the same system used across Google web apps. Each operation maps to an obfuscated RPC method ID:

All 9 artifact types (except Mind Map) use the unified CREATE_ARTIFACT RPC with different type codes. The plugin handles response parsing, error detection, and the anti-XSSI response format automatically.

Cookie Management

Cookies are stored in a platform-aware location:

The plugin captures all Google domain cookies (30+) via Chrome DevTools Protocol — this is required because Google's passive login check validates cookies beyond the standard auth set.

Known Limitations

• Audio / Video / Slides download: These streaming media types are created successfully on NotebookLM's servers, but auto-download requires browser-level cookie handling. The plugin returns the artifact URL for manual browser download. Static media (infographics, reports) download automatically.
• Data Table: Parameter structure for type 9 artifacts is still being reverse-engineered.
• RPC method IDs: Google can change these at any time. If generation fails, check for updated IDs in the notebooklm-sdk project.

🌟 Star History

💜 Sponsors

This project is free and open source. Sponsors help keep it maintained and growing.

Become a Sponsor | Sponsorship Program

🤝 Contribution

When contributing to this repository, please first discuss the change you wish to make via issue,
email, or any other method with the owners of this repository before making a change.

Contributions are what make the open-source community such an amazing place to learn, inspire, and create. Any contributions you make are greatly appreciated ❤️.

You can learn more about how you can contribute to this project in the contribution guide.

⚖️ License

This repository is available under the MIT License.

👍 Credits

• notebooklm-sdk — TypeScript SDK reference for batchexecute protocol
• notebooklm-kit — Artifact creation patterns
• notebooklm-py — Python RPC reference implementation
• sherlock-ai-plugin — CDP authentication patterns

Happy vibe coding 💯

Made with ❤️ by Proyecto 26 - Changing the world with small contributions.

One hand can accomplish great things, but many can take you into space and beyond! 🌌

Together we do more, together we are more ❤️