On day three, I attempted to develop a skill in Claude Code. As I understand it, we can build skills for the tasks we perform regularly and then create automations later. So, I’m planning to create an automation to generate SEO reports in the future. I uploaded the provided files, skill.md and reference.md, into my VS Code and asked Claude to create a skill tailored to my needs. I brainstormed with the plan mode to develop it, and after executing the skill, I ran a demo report using dummy data, and it successfully produced a report. Of course, it needs refinement in the future by connecting to real APIs and tools, but given the successful first demo report, I’m pleased.

On day two, I started mindlessly doing everything without reading the Calude code. However, I soon realised I was encountering numerous errors. I noticed I lacked a basic setup on my computer like installing Node.js. Consequently, I couldn’t connect the Firecrawl MCP in my environment. I sorted this out and everything worked smoothly. Next, I attempted to scrape a competitor’s website to gather some important information for later use in building my own website.

Hello everyone, I've completed my first task of building a newsletter using Claude code. Since it was my first time working with Claude code, I wanted to keep things simple by following the first lesson by Nate. Everything executed perfectly without any errors for this task. For the second day, I plan to try something on my own while still following the lessons for the upcoming days. I would really appreciate it if anyone could share their use cases. Thank you!

Like most people, I'd been using ChatGPT to write things, ask questions and review the odd bit of code, and I thought that meant I was using AI. It didn't. Not even close. Then I stumbled onto Nate's channel (and now this community), and it honestly changed the way we work. The WAT framework was probably the biggest lightbulb moment. Nate explains agentic workflows in a way that just makes sense. I watched the 36-minute Claude Code beginner video three times. Not because I had to, but because every time I watched it, something else clicked. Since then we've completely changed how we build software. We now have agents with specific jobs, memory that carries context between sessions, reusable skills, and workflows that mean we're not constantly repeating ourselves. Instead of spending hours on the boring repetitive stuff, we let the agents handle it and focus on the parts that actually need us. The time savings have been insane. Projects we'd normally estimate in months are now taking weeks. But the biggest change wasn't even the speed. I actually enjoy working again. There were parts of this job I used to dread because they were repetitive or just mentally exhausting. Now I catch myself looking forward to opening my laptop in the morning. I honestly didn't expect AI to do that. Massive thanks to Nate for everything he shares. And for those just checking in, wondering whether it's worth learning... it is.

Completed day 2 and glad I finally got to see what Firecrawl can do. I always heard it can scrape sites but hadn't seen any use cases. Had a hiccup along the way with connecting to firecrawl's MCP server from Claude Code… think in the video Nate used the 'claude code' section command but that didn't work for me. It worked when I used the 'VS Studio' link on the Firecrawl website to connect to MCP (maybe it's changed since the video was made). 2nd small hiccup was when my Firecrawl account ran out of credits while Claude Code was running the prompt to scrape a job site to a csv file. It just stopped with less listings so I still got to see the final product which was cool (this task took 1000+ credits 🤷‍♀️). I used my work email to sign up for more credits to run the next use cases which only took 1 credit each haha. Both created nice looking artifacts below, analyzing Myclaw's brand colors and Blue Bottle's site mapping which was impressive!