Building a Cricket Scraper for My Obsidian Vault feat. Google Antigravity
Building CricketCLI with Google Antigravity took an hour. It was impressively fast and strangely hollow. Here's what I learned about AI-assisted development.
With my CLI tools, I was able to create MD files with proper frontmatter for movies, video games, books, and TV shows. All of these media had proper APIs that presented me with the data I required.
But sports, especially cricket? Finding an API that presented me with relevant information was hard. Either the APIs were paywalled or nothing robust existed.
This issue was on my back burner but kept eating at me. So when Gemini 3 and Google Antigravity were released, I wanted to test them out.
You can view the final code on GitHub before reading how I built it.
So I downloaded Antigravity, and opened it up to find it very much to be built on top of VS code. But the hook was that, you have agents also controlling it. I used Gemini 3 low under the hood but you can choose your LLM of choice.
I told the agent what I wanted to do, it created a Implementation Plan markdown, as well as Tasks markdown. Then step by step, from analyzing the Cricbuzz HTML structure to testing by creating a dummy file, it built the tool.
I was not happy with the CLI UX, so I asked the agent to add a Fuzzy search and it complied.
The whole affair took less than an hour. Writing a web scraper was something I don't have a lot of experience with, nor does it particularly interest me. So that was the most impressive part for me, watching the agent handle the exact thing I would've dragged my feet on.
But as impressive as it sounds, there was a feeling of emptiness. I like getting my hands dirty, to be in the zone or flow. That feeling when I'm coding, creating, developing—it seemed to have been lost.
It's not like I haven't used coding agents before. But there was a sense of being an architect, where I was still assembling things, figuring out the flow, and revising the code. That joy seems to have been lost here.
The world is quickly moving in this direction, and for me to dwell on these grievances means I'll be left behind. The smarter thing to do is to adapt and move on to solving a different set of problems.
Even for fun, there will always be problems to solve—like figuring out how to create agents that can help me with my productivity or note-keeping. If not that, maybe something else that will scratch the itch. And honestly, I never would have written a web scraper any other way.
If you'd like to follow my journey, please subscribe. If you have thoughts or suggestions, feel free to leave a comment or reach out to me.