notebooklm-skill 这个技能将 NotebookLM 的调研能力，与 Claude Code 结合来起来。

只需扔给它一批网址、PDF 或者热点话题，就能自动创建笔记、做研究，整理成结构化的信息。

再拿着这份内容，给 Claude 接着写文章、发推文、Newsletter，想要什么文本格式都可以。

GitHub：http://github.com/claude-world/notebooklm-skill…

除了能当 Claude Code 的 Skill 用，也能当独立的 MCP 服务跑，接进自己 Agent 工作流。

还能顺手生成播客音频、幻灯片、思维导图这些，能帮做自媒体的朋友节省不少时间。

claude-world/notebooklm-skill

Source: https://github.com/claude-world/notebooklm-skill

notebooklm-skill

NotebookLM does the research, Claude writes the content.

The only tool that connects trending topic discovery → NotebookLM deep research → AI content creation → multi-platform publishing. Works as a Claude Code Skill or standalone MCP Server.

繁體中文版 README

Demo

All slides, podcasts, and videos were generated by NotebookLM using this tool.

What is this?

notebooklm-skill bridges NotebookLM’s research capabilities with Claude’s content generation. Feed it URLs, PDFs, or trending topics — it creates a NotebookLM notebook, runs deep research queries, and hands structured findings to Claude for polished output: articles, social posts, newsletters, podcasts, or any format you need.

Built on notebooklm-py v0.3.4 — pure async Python, no OAuth setup needed.

Sources (URLs, PDFs)          NotebookLM                Claude               Artifacts & Platforms
+-----------------+    +------------------+    +-----------------+    +----------------------+
| Web articles    |--->| Create notebook  |--->| Draft article   |--->| Blog / CMS           |
| Research papers |    | Add sources      |    | Social posts    |    | Threads / X          |
| YouTube videos  |    | Ask questions    |    | Newsletter      |    | Newsletter           |
| Trending topics |    | Extract insights |    | Any format      |    | Any platform         |
+-----------------+    +------------------+    +-----------------+    +----------------------+
     Phase 1                Phase 2                Phase 3                  Phase 4
                                |
                                v
                       +------------------+
                       | Generate artifacts|
                       | Audio (podcast)   |
                       | Video             |
                       | Slides            |
                       | Report            |
                       | Quiz              |
                       | Flashcards        |
                       | Mind map          |
                       | Infographic       |
                       | Data table        |
                       | Study guide       |
                       +------------------+
                            Phase 2b

Quick Start

# Option A: uvx (recommended — zero install)
uvx notebooklm-skill --help
uvx --from notebooklm-skill notebooklm-mcp   # Start MCP server

# Option B: pip install from PyPI
pip install notebooklm-skill

# Option C: Install from source
git clone https://github.com/claude-world/notebooklm-skill.git
cd notebooklm-skill && pip install .

# Option D: One-line install (pip + Playwright + Claude Code Skill)
git clone https://github.com/claude-world/notebooklm-skill.git
cd notebooklm-skill && ./install.sh

# Authenticate with Google (one-time, opens browser)
uvx notebooklm login              # if using uvx
# or: python3 -m notebooklm login  # if using pip install

# Use commands (uvx or direct — both work the same)
notebooklm-skill create --title "My Research" --sources https://example.com/article
notebooklm-skill ask --notebook "My Research" --query "What are the key findings?"
notebooklm-skill podcast --notebook "My Research" --lang en --output podcast.m4a
notebooklm-pipeline research-to-article --sources https://example.com --title "Topic"
notebooklm-mcp                   # Start MCP server (stdio mode)

Or use scripts directly: python scripts/notebooklm_client.py create ...

See docs/SETUP.md for the full setup guide.

Authentication

notebooklm-py uses browser-based Google login. No API keys, no OAuth Client ID, no Google Cloud project needed.

# One-time login (opens Chromium, sign in with Google)
uvx notebooklm login              # if using uvx
python3 -m notebooklm login       # if using pip install

Session typically lasts weeks. Re-run login if you get authentication errors.

Two Ways to Use

Features

Artifact Generation (9 downloadable types)

Most artifacts support language selection (e.g., --lang zh-TW). Exceptions: quiz, flashcards, mind-map.

Note: NotebookLM returns audio in MPEG-4 (M4A) format, not MP3.

Architecture

+---------------------------------------------------------------+
|                      notebooklm-skill                          |
|                                                                |
|  +---------+  +--------------+  +----------+  +------------+  |
|  | Phase 1 |  |   Phase 2    |  |  Phase 3 |  |  Phase 4   |  |
|  | Collect  |->|  Research    |->| Generate  |->|  Publish   |  |
|  +---------+  +--------------+  +----------+  +------------+  |
|      |              |                |               |         |
|  +--------+  +-------------+  +-----------+  +-----------+    |
|  | URLs   |  | NotebookLM  |  |  Claude    |  | Threads   |    |
|  | PDFs   |  | (via        |  |  Content   |  | Blog      |    |
|  | RSS    |  |  notebooklm |  |  Engine    |  | Email     |    |
|  | Trends |  |  -py 0.3.4) |  |            |  | CMS       |    |
|  +--------+  | - notebooks |  +-----------+  +-----------+    |
|              | - sources   |        |                          |
|              | - chat/ask  |  +-----------+                    |
|              | - artifacts |  | Artifacts |                    |
|              +-------------+  | audio     |                    |
|                               | video     |                    |
|                               | slides    |                    |
|                               | report    |                    |
|                               | quiz      |                    |
|                               | flashcards|                    |
|                               | mind-map  |                    |
|                               | infographic| ⚠️ no download    |
|                               | data-table|                    |
|                               | study-guide|                   |
|                               +-----------+                    |
|                                                                |
|  +-----------------------------------------------------------+ |
|  |  Interfaces                                                | |
|  |  +-- scripts/          CLI tools (notebooklm-py direct)   | |
|  |  +-- mcp_server/       MCP protocol server                 | |
|  |  +-- SKILL.md          Claude Code skill definition        | |
|  +-----------------------------------------------------------+ |
+---------------------------------------------------------------+
         ^                                          ^
         |                                          |
   +-----------+                             +-----------+
   |trend-pulse|                             |threads-   |
   |(optional) |                             |viral-agent|
   +-----------+                             |(optional) |
                                             +-----------+

Usage Examples

1. Research to Article

python scripts/pipeline.py research-to-article \
  --sources "https://arxiv.org/abs/2401.00001" \
            "https://blog.example.com/ai-agents" \
  --title "AI Agent Survey"

2. Research to Social Posts

python scripts/pipeline.py research-to-social \
  --sources "https://example.com/ai-news" \
  --platform threads \
  --title "AI News This Week"

3. Trending Topics to Content

python scripts/pipeline.py trend-to-content \
  --geo TW \
  --count 5 \
  --platform threads

4. RSS Batch Digest

python scripts/pipeline.py batch-digest \
  --rss "https://example.com/feed.xml" \
  --title "Weekly AI Digest"

5. Generate All Artifacts

python scripts/pipeline.py generate-all \
  --sources "https://example.com/article" \
  --title "Research" \
  --output-dir ./output \
  --language zh-TW

6. Slides + Podcast → YouTube Video

Combine NotebookLM-generated slides and podcast into a YouTube-ready video:

# Generate slides and podcast
python scripts/notebooklm_client.py generate --notebook "Research" --type slides
python scripts/notebooklm_client.py podcast --notebook "Research" --lang en --output podcast.m4a
python scripts/notebooklm_client.py download --notebook "Research" --type slides --output slides.pdf

# Convert PDF to PNG + compose video
./scripts/make_video.sh slides.pdf podcast.m4a output.mp4

Pipeline Workflows

MCP Server Setup

Add to your project’s .mcp.json:

{
  "mcpServers": {
    "notebooklm": {
      "command": "uvx",
      "args": ["--from", "notebooklm-skill", "notebooklm-mcp"]
    }
  }
}

Or if you installed via pip install notebooklm-skill:

{
  "mcpServers": {
    "notebooklm": {
      "command": "notebooklm-mcp"
    }
  }
}

Works with Claude Code, Cursor, Gemini CLI, and any MCP-compatible client.

Claude Code Skill Setup

# Option A: Symlink (auto-updates with git pull)
./install.sh

# Option B: Manual copy
mkdir -p .claude/skills/notebooklm
cp /path/to/notebooklm-skill/SKILL.md .claude/skills/notebooklm/
cp /path/to/notebooklm-skill/scripts/*.py .claude/skills/notebooklm/scripts/
cp /path/to/notebooklm-skill/requirements.txt .claude/skills/notebooklm/

# Authenticate (one-time)
python3 -m notebooklm login

Claude will automatically detect the skill when you ask about research, NotebookLM, or content creation.

API Reference

CLI Commands (11)

MCP Tools (13)

Integrations

• trend-pulse — Real-time trending topic discovery from 7 sources
• threads-viral-agent — Auto-publish research-backed social posts

Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes
4. Push and open a Pull Request

# Development setup
git clone https://github.com/claude-world/notebooklm-skill.git
cd notebooklm-skill
pip install -e .
python3 -m notebooklm login
python -m pytest tests/

License

MIT License. See LICENSE.