Type a username, get every repo
Published: 8/8/2026
The GitGenius lookup box used to want owner/repo and nothing else. Type torvalds and you got "Repository not found on GitHub.com," which is technically true and completely unhelpful — that isn't a typo, it's a person telling us what they're interested in using fewer characters than we asked for.
A large share of what people typed into that box was a bare username. Now it works.
What the box accepts
Type any of these and we'll figure out what you meant:
facebook/reacthttps://github.com/facebook/reactgithub.com/facebook/reactfacebook
The classifier tells users from repositories by grammar rather than by guessing. GitHub logins are alphanumerics with single interior hyphens, capped at 39 characters — the same rule for users and organizations. Repository names are far more permissive, allowing dots and underscores too. One path segment that parses as a login is a user; two segments are an owner and a repo. Anything else comes back as invalid, with an error that says so instead of blaming GitHub.
What you get back
A list of that account's public repositories, newest activity first, with each one labeled by whether GitGenius already tracks it.
That label is the part that matters. The list mixes repos with full history, charts, and issue analytics ready to look at, and repos we've never seen. Both are useful — one you can open right now, the other is one click from being added — but they're different actions, and a list that didn't distinguish them would send you to a dozen empty pages to find out which was which.
Above eight repositories you get filter chips: all, indexed, not indexed. Prolific accounts paginate at twenty. Each row shows how long ago it was last updated, so a list of sixty repos doesn't read as sixty equally live projects.
Adding what's missing
Every unindexed repo in the list has its own add button. No copying the name into a second form, no leaving the page.
Worth setting expectations: adding a repo starts the indexing process rather than finishing it. History has to be pulled and analytics computed before charts and issue summaries mean anything, so a freshly added repo will show a page that fills in over time rather than one that's complete on arrival.
Organizations work the same way as users here — kubernetes and torvalds both resolve to a repository list. Note that this is about listing an org's repositories; org-level indexing is separate and isn't live yet.
Why this was worth doing
This is a small change. It's also the difference between a first session that produces something and one that produces an error message.
The failure mode it removes is the worst kind: the user did nothing wrong, the system had everything it needed to help, and it said no anyway because the input didn't match a format nobody told them about. Every lookup box eventually accumulates a version of that. This one had a particularly common one.
It's live on the homepage and on the repository lookup page. Try your own username.