That's great, will you also add the category field to the text file in the repo? These projects are often heavy on tech blogs and I'd like to filter those out.

We are classifying them at runtime, per entry. More accurate and LLMs are cheap.