#7dayAISChallenge - Day 2 Build: Course Scraper & Auto-Downloader! 📚

Hey everyone! 👋 Day 2 is officially in the books and the capabilities we are unlocking with MCPs are seriously exciting.

🛠️ What I built / scraped:

I used the Firecrawl MCP to build a web scraper for an online Generative AI course. It pulls the tables containing all the course materials and homework, outputs that structured data into both a Google Sheet and an Excel file, and then automatically downloads all the PDF slides for me. (I’m attaching screenshots of my Claude Code project setup and the final output below!)

🧠 One thing I learned:

MCPs absolutely supercharge what Claude Code can do. But my biggest "aha" moment was regarding context management: I learned that preparing a reference file before creating the claude.md file is a massive time and context-saver and keeps the agent much more focused.

❓ Question for the advanced builders:

I’m a bit confused about one architectural choice. For this specific project, I didn't use the WAT framework for my claude.md. Instead, I just asked Claude Code to generate a fresh claude.md directly based on my request.

When exactly should we be strictly adhering to the WAT framework for claude.md, versus just asking Claude Code to generate a fresh one from scratch? I’d love to hear how you all approach this decision!

🚀 One thing I'd improve (My next step):

After I stress-test this workflow on a couple of other online educational platforms to make sure it's robust, my goal is to package this entire automation into a reusable Skill.

Can't wait to see what everyone else is scraping today! Let's keep building! 🔥

2 comments