How BestOfGit works
BestOfGit tracks open-source repositories and shows which ones are gaining attention right now. It reads public GitHub data and never asks you to sign in.
Where the data comes from
Four sources, all of them public:
- GitHub's own trending pages, for the daily, weekly, and monthly views.
- The GitHub Search API, which widens the net beyond trending so that a repository can be picked up before it is popular.
- A second pass over the GitHub Search API for repositories that were popular once and have gone quiet: 5,000 stars or more, and no push for eighteen months.
- Our own daily snapshot of every tracked repository's star count. This is what makes it possible to talk about change rather than just totals.
How repositories are ranked
The ranking is the change in star count over the selected window, not the total number of stars. A repository that gained 300 stars this week ranks above one that has 80,000 stars and gained nothing. The window defaults to the last 7 days and can be switched to 24 hours, 30 days, or a year.
Star counts are sampled once a day, so each window needs two snapshots far enough apart: the 24-hour view works from day 2, the 7-day view from day 8, the 30-day view from day 31, and the year view from day 366. Until then those windows stay empty rather than being filled with guesses.
Repositories are also connected to each other by the topics they carry: every project page lists the repositories that share at least two of its topics, and the statistics page shows which topic pairs appear together most often.
Why some repositories are hidden
Not everything on GitHub is a software project: a repository of personal notes, a link dump, or a configuration mirror can be popular without belonging in a list of projects. Two filters keep those out of the default views: a set of heuristics that reject obvious cases cheaply, and a classifier that judges the rest. Nothing is deleted: every repository that was rejected, and every one the classifier has not judged yet, is listed with its reason on the hidden page.
How fresh the numbers are
The pipeline runs every six hours in several stages: discovery, star snapshots, scoring, and classification. When a stage stops reporting, the numbers here are still the last ones we have, and the site says so at the top of every page.
Good to know
- Every timestamp on this site is UTC.
- Star counts come from the GitHub API: read by our pipeline, never by your browser.
- BestOfGit is an independent project and is not affiliated with GitHub.
Legal and data
The code is open source (MIT) and the data belongs to the people who made the repositories. The terms of use cover what you may do with what you find here, how repository owners can ask for a correction or removal, and what we do not promise about the numbers. The privacy page describes what is and is not processed when you visit: no accounts, no cookies, no third-party analytics.