What problem does it solve? When an executive needs to learn or synthesize a topic, ad-hoc web searches produce inconsistent, low-quality results. This Skill standardizes how agents use NotebookLM as a knowledge engine: it enforces a CLI-first routing policy, curates authoritative sources when none are provided, and falls back to deep research or web search when documentary gaps appear. ## Core Features & Use Cases - CLI-first routing with MCP fallback: Prefers the nlm CLI (version >= 0.7.0) and uses notebooklm-mcp only when the CLI route is unavailable, with explicit authentication validation via nlm login --check. - Automatic source curation: When no sources are provided, collects the top 10 sources ranked by authority, recency, coverage, diversity, and traceability, rejecting SEO spam and duplicates. - Gap-driven deep research: Detects documentary gaps after the main query and triggers deep research or web search, then re-queries the notebook. - Use Case: Ask to learn about 'AI regulation in the European Union' — the agent validates the runtime, creates or resolves a notebook, ingests the 10 best sources, synthesizes an answer, and delivers the summary with a transparent source list. ## Quick Start Use NotebookLM to research and synthesize the best current sources on EU AI regulation, listing every source you used.
