Content Blueprint
A fully autonomous AI research agent powered by Google NotebookLM that transforms trending topics and source documents into polished, publishable content across multiple formats.
Run this command to deploy the blueprint to your environment.
This skill ingests sources into NotebookLM, runs cited research queries, generates native artifacts (podcasts, videos, reports, quizzes, slides, flashcards, infographics, data tables, and study guides), and synthesizes research findings into original content using Claude. It eliminates the manual work of organizing disparate sources, extracting key findings, and translating research into multiple content formats.
Once authenticated with Google (one-time browser login), you provide source URLs, PDFs, or topics. The skill creates a NotebookLM notebook, ingests sources, and either generates downloadable artifacts directly (audio, video, slides, reports) or uses cited research answers to power Claude-written articles, social posts, and newsletters. For trending topics, it auto-discovers sources and drafts content. Output depends on your choice: native NotebookLM artifacts (podcast MP4, slide deck PDF, quiz JSON) or Claude-generated content (Markdown articles, platform-specific social posts).
Ingest sources into NotebookLM, run cited research, and generate podcasts, videos, reports, quizzes, slides, flashcards, infographics, and data tables. Also use research to power Claude-written articles and social posts.
Research-backed content that cites sources automatically. Every claim includes the source passage and exact location, so you publish with confidence instead of fact-checking manually.
Requires Python 3.10+ because the skill uses native asyncio for concurrent API calls. Authenticates via browser login to NotebookLM (no API keys needed) and stores the session locally. Built on notebooklm-py v0.3.4, a pure async Python client with 8 sub-APIs covering notebooks, sources, artifacts, chat, research, notes, settings, and sharing. The MCP server uses FastMCP for MCP-compatible clients (Claude Code, Cursor, Gemini CLI).
Audio and video generation take 3-15 minutes and require patience on the first run. Infographic downloads are unreliable (use slides instead). Session tokens expire after weeks. Rate limits are undocumented but roughly 5-10 artifact generations per hour. Requires Python 3.10+ and a Google account with NotebookLM access.
Source Ingestion (8 types)
Ingest URLs, PDFs, YouTube videos, Google Drive docs, plain text, Markdown, DOCX, and CSV files into a single research notebook.
Cited Q&A
Ask questions and get answers with automatic source citations, passage references, and follow-up conversation support.
9 Downloadable Artifacts
Generate podcasts (M4A), videos (MP4), slide decks (PDF), reports (Markdown), quizzes, flashcards, mind maps, infographics, and data tables in native NotebookLM quality.
Deep Web Research
Automatically discover relevant sources on the web or in Google Drive, then import the top findings into your notebook to deepen analysis.
Multi-Format Content
Combine NotebookLM research with Claude to write articles, social posts, newsletters, and custom reports tailored to your voice.