The library stopped getting commits. A fork didn't.
We crawl GitHub for repositories that have gone quiet, then surface the forks that are commits ahead and still shipping. 193,000 repositories and 2.2 million forks tracked so far.
Most widely-used repositories have forks that are ahead of the original — sometimes a community keeping an abandoned project alive, sometimes a maintainer who took it somewhere new. Both are worth knowing about before you vendor a dead dependency or write the patch yourself.
12,000 developers have signed up.
Every two weeks No drip campaign, no upsells
10 repositories Chosen by fork activity, not stars
1 to 5 forks each With commits ahead, behind, and last commit date
Fresh out of the crawler
Unedited, straight from the database. Each row is a repository we track and a fork of it that is carrying commits the original doesn't have. The GitHub links go to both, so you can check every number yourself.
-
prefix-dev/pixi 7.6k ★
Powerful system-level package manager for Linux, macOS and Windows written in Rust – building on top of the Conda ecosystem.
-
stablyai/orca 46.5k ★
Orca is the ADE for working with a fleet of parallel agents. Run any coding agent with your own subscription. Available on desktop, mobile and VPS.
-
oss-apps/split-pro 1.4k ★
Open source alternative to Splitwise
-
lupinemachines/lupine 2.4k ★
LUPINE is a GPU over IP bridge allowing GPUs on remote machines to be attached to CPU-only machines.
How it works
1. Crawl
A worker walks GitHub through its API and records every fork of every repository we track, along with each fork's commit history. 2.2 million forks so far.
2. Compare
Each fork is diffed against its parent: commits ahead, commits behind, date of last commit. The case we're looking for is a fork moving forward while the parent has stopped.
3. Digest
Every two weeks, the ten strongest examples go out by email with the numbers attached, so you can judge whether a fork is worth switching to rather than taking our word for it.
What lands in your inbox
Plain HTML, no tracking pixels, no images to load. Each entry is one repository, its last commit date, and the forks ahead of it with their counts.
Read one before you subscribe
Every issue since 2020 is public — no email required. If the picks aren't useful to you, you'll know in two minutes instead of two weeks.
Browse the archiveCheck a specific repository
Paste any GitHub repo and see its forks ranked by commits ahead, with the last commit date on each. 193,000 repositories are already in the database; anything we haven't seen gets queued for the crawler.
Check a repositoryOr query it from your terminal
The same data is available over HTTP. No key, no signup.
curl https://activeforks.net/api/1.0/ActiveForks/expressjs/body-parserQuestions
Why would I want a fork instead of the original?
Because the original may have stopped moving while the work didn't. When a library goes quiet, the fixes usually don't stop existing — they move into somebody's fork: a security patch, support for the current runtime version, the PR the maintainer never merged. Your options with a dead dependency are to vendor it, patch it yourself, or find whoever already did the work. This site is for the third one. It also catches the reverse case: a maintainer who restarted the project in a fork and never redirected the old repo, so the URL everyone links to is the dead one.
How do repositories get picked?
Two ways in. A worker searches GitHub for repositories with 200+ stars pushed in the last 30 days, in star buckets, so nothing widely used slips past the search API's result cap. Anything you look up on the check-a-repo page or through the API also gets queued and crawled. For each repository we take the most-starred forks and run a commit comparison on every one that has been pushed since it was created. Only forks that are genuinely ahead of the parent get stored. No human curation and no sponsored picks — a query decides, and every past issue is public, so you can audit its taste before trusting it.
What do "ahead" and "behind" actually mean?
They're GitHub's own comparison numbers for parent versus fork. Ahead is commits in the fork that the parent doesn't have. Behind is commits in the parent that the fork hasn't pulled in. A fork 40 ahead and 0 behind has taken the project forward and missed nothing; 40 ahead and 900 behind diverged years ago and stopped following along. What the numbers can't tell you is what's in those commits — treat them as a shortlist, not a verdict, and click through to the diff.
Is it free?
Yes. No account, no paid tier, no ads, no sponsored picks. The emails carry no tracking pixels and no images to load. Browsing the archive and calling the API cost nothing; subscribing costs one email address.
How current is the data?
The crawler runs continuously, but it won't re-check the same repository more than once every 30 days — GitHub's rate limits make a faster full sweep impossible, and fork activity doesn't change hour to hour. So the commit counts you see are from our most recent crawl of that repository. When a number has to be exact, the GitHub links next to it are authoritative.
Can I check my own repository?
Yes. Put owner/name into check a repo.
If it's already in the database you get its active forks immediately, ranked by
commits ahead. If we've never seen it, your lookup puts it in the crawl queue —
come back in a few minutes. Two limits: the repository has to be public, and it has
to be an original rather than itself a fork, since the whole comparison is
fork-against-parent.
Who runs this? Is it affiliated with GitHub?
One developer, Floatas, since 2020. Not affiliated with, endorsed by, or funded by GitHub; it reads the public API like any other client. Corrections, or a repository you think we've got wrong: floatas [ eta ] outlook . com
What happens after I subscribe?
You get a confirmation link first — nothing is sent until you click it. After that, one email every two weeks. Every issue has an unsubscribe link, and the list is never sold, shared, or used for anything but the newsletter.