Public repositories on Weft

Browsing returns every public repository in namespace order, alphabetically. That is the whole rule. It is not a quality signal, nobody is promoted into it, and there is no placement to buy.

Searching matches a case-insensitive substring of the name, the namespace, the description or any of the repository’s topics. Results come back in the same alphabetical order — there is no relevance score, so a repository is not ahead of another because an algorithm liked it more.

New & notable is the most recently created repositories among the first few hundred this page has loaded. The API has no date ordering yet, so that window is real and stated rather than dressed up as a global top list.

Topics are real; languages are not, yet. The topic chips are the topics maintainers have actually filed things under, most-used first, with the number of repositories carrying each — read from the same endpoint the repository pages write to, and scoped to what you may see, so a topic used only inside a private namespace is not listed to a stranger. Clicking one filters by the topic exactly, not by the word: a repository appears under kubernetes because somebody filed it there, never because its description says so. The language chips are still plain text searches: Weft does compute a repository’s language mix and shows it on the repository page, but search cannot filter by it, so a language chip finds repositories that say the word rather than ones written in it. Said here rather than left to be discovered.

There is no trending tab. Ranking a fortnight of traffic produces a random number wearing a confident face. It arrives when there is a month of signal behind it, and until then this page would rather be boring than wrong.

Private repositories are never indexed, never suggested and never counted here, whether or not you are signed in.