more about WW extension project

How it has gone

Here is an update about the project process.

As I expected the implementation was really quick. I could build it with claude code only within two days.

First day for version 1, which is mainly about the key feature the job scaning and scoreing. Second day was for version 2, which is about User customization and token optimization. this was the most of the project, and after that, I worked on Version 3, which is about documentizing and UX upgrade.

Currently, I’m waiting for Chrome Webstore draft review, to post it on the webstore. This is only for Uwaterloo students but I did used this tool really usefully, so just wanted to share it.

What did I learned?

I can say this project was the best done AI project, because I felt I had a full control over the agents and whole project.

this might be becuase the size of project is small and didn’t use technology I never used before. But regardless to those, I felt more control over it.

The key difference between this project and others like empkl or macro-dashboard was the harness engineering.

I’m not sure Harness engineering is the right word, but I did make some setup about agent’s behavior before even explaining what the project will be.

after those documentization of that rules and structures, I followed the way I used to work with claude code.

With this simple process of building harness, the developing experience increased so much, so I finally knew the harness engineering is real.