1. How are the repos discovered and maintained?

  • (Daily) I have a script that searches through >100 keywords and topics to discover new repos.
    • Repos are annotated by AI (gpt-5.2-mini). Its agreement with my annotations is ~75%, but most of the disagreements are the repos that I myself am uncertain how to categorize.
  • (Daily) I have another script that updates existing repos.
  • (Once in a while) I have another script that uses AI to annotate locations for repos and developers.

2. Why do you only show repos with >= 500 stars?

GitHub rate limits. I'm using free GitHub API so I can only query a certain number of repos an hour. Reducing the number of minimum stars will make the number of repos I need to query too much.

If you want to sponsor the project to switch to paid API, email me!

3. Is star diff a good indication for good repos?

The goal of the project is to help you discover trending repos. Evaluating a repo is up to you. What I've found that if a repo is no longer maintained or not super useful, its growth will taper off quickly.

4. What insights did you get from looking at so many repos?

I like doing data analysis, and here is what I've learned from analyzing these repos in the past.

To be updated about my future analysis, check out my substack, X, or LinkedIn.

If you want to collaborate on market research, email me!