# NotebookLM Assistant Skill
This skill enables automated content processing and generation through NotebookLM, converting various content sources into multiple output formats.
## Supported Content Sources
| Source Type | Description |
|-------------|-------------|
| Web Pages | Any URL or web article |
| YouTube Videos | Video transcripts and content |
| Office Documents | Word (.docx), PowerPoint (.pptx), Excel (.xlsx) |
| PDFs | PDF documents |
| eBooks | EPUB files |
| Images | With OCR text extraction |
| Audio Files | With transcription |
| Structured Data | CSV, JSON, XML files |
| Plain Text | Markdown (.md), Text (.txt) |
| ZIP Archives | Compressed file bundles |
## Output Formats
Generate content in various formats using natural language commands:
| Format | Example Commands |
|--------|------------------|
| Podcast | "generate podcast", "create audio discussion" |
| Video | "make a video", "generate video content" |
| PPT Slides | "create slides", "make a presentation", "generate PPT" |
| Mind Map | "create mind map", "generate concept map" |
| Infographic | "make infographic", "create visual summary" |
| Quiz | "generate quiz", "create test questions" |
| Flashcards | "make flashcards", "create study cards" |
| Report | "write report", "generate summary report" |
| Data Table | "create table", "organize as data table" |
## Setup Requirements
### 1. NotebookLM Authentication
Before first use, authenticate with NotebookLM:
```bash
notebooklm login
```
Verify authentication:
```bash
notebooklm list
```
### 2. MCP Server Configuration (Optional)
For extended content source support, configure MCP servers in `~/.claude/config.json`:
```json
{
"mcpServers": {
"notebooklm": {
"command": "notebooklm",
"args": ["mcp"]
}
}
}
```
Restart Claude after configuration changes.
## Workflow
1. **Content Detection**: Automatically identify the content source type from user input
2. **Content Extraction**: Retrieve and process content using appropriate tools
- Direct URL fetch for web pages
- YouTube transcript extraction for videos
- Document parsing for Office files/PDFs
- OCR for images
- Transcription for audio
3. **Upload to NotebookLM**: Add processed content as notebook source
4. **Generation**: Create requested output format based on user intent
5. **Download**: Save generated outputs to local storage
## Usage Examples
### Generate a podcast from a web article
```
Create a podcast from https://example.com/article
```
### Create slides from a PDF
```
Make a presentation from report.pdf
```
### Generate a mind map from YouTube video
```
Create a mind map from https://youtube.com/watch?v=xxxxx
```
### Create flashcards from a document
```
Generate flashcards from study-guide.docx
```
### Make an infographic from structured data
```
Create an infographic from data.csv
```
## Error Handling
- If authentication fails, run `notebooklm login` again
- For unsupported file types, convert to a supported format first
- Large files may require chunking or compression
- Network errors will prompt automatic retry
## Notes
- Processing time varies based on content size and output type
- Some outputs (like video) may take longer to generate
- Generated files are saved to the current working directory by default
