Feb 28
Loading video player...
In this video I show you exactly how I build Claude Code agent skills from scratch using NotebookLM — now powered by Gemini 3.1 with deep research. This is the workflow I've been using to create custom skills for product management, third-party API integrations, and everything in between. The process is simple: use NotebookLM to gather as many sources as possible on your topic, then prompt it to generate a structured agent skill file. Load that skill into Claude Code and you now have a deeply informed AI agent that understands your specific domain — whether that's the Hooked model for product retention, the Stripe API, Convex, or any third-party integration your client needs. I also walk through a live demo using Harbor SEO.AI as the test subject — feeding it a product management skill and watching it identify exactly the right retention gaps (variable rewards, re-engagement timing, performance notification loops) with zero context. The quality of suggestions from a NotebookLM-generated skill is genuinely unbelievable. Key things you'll learn: → How to build Claude Code skills in 1-2 minutes using NotebookLM → Why Gemini 3.1 + deep research makes NotebookLM dramatically better → How to use this for third-party API integrations (Stripe, Convex, Guesty, etc.) → Real example: product management skill applied to a live SaaS app Try HarborSEO FREE: https://harborseo.ai Work with us: https://incomestreamsurfer.com Try Bright Data: https://brightdata.com/?promo=incomestreamsurfers 0:00 How I build Claude Code skills with NotebookLM 0:15 Creating a new notebook — habit forming apps 1:00 Adding the Claude Code skills page as a source 1:25 Getting more sources with fast research 1:55 Why I'm obsessed with the Hook Model (Nir Eyal) 2:45 Harbor SEO.AI — how we applied the Hook Model 4:05 Generating the agent skill file in NotebookLM 4:35 Harbor SEO going paid in 48 hours 5:36 Loading the skill into Claude Code 6:15 NotebookLM now runs on Gemini 3.1 + deep research 6:50 Running the skill — product analysis live 7:30 Using NotebookLM for third-party API skills (Guesty example) 8:45 Why this is better than feeding docs manually 9:22 Results — what the skill found about Harbor 12:55 Wrap up #ClaudeCode #NotebookLM #Gemini31 #AIcoding #ClaudeCodeSkills #VibeCoding #NotebookLMTutorial #AITools #ClaudeAI #AICodingTools