Downloads come from GitHub's release API. Installs come from the app checking in at most once a day. Everything here is a real count of a real thing.
One bar per day. A device counts once, however often it opens the app.
Reported by the app itself at check-in.
API level, as reported by each device.
Model strings. These identify a handset model, not a person.
Straight from the GitHub Releases API.
| Release | Published | Downloads |
|---|---|---|
| loading | ||
No invented numbers, and no rounding up either.
Read from the GitHub Releases API, which counts every fetch of the APK asset. It does not count a download that was later deleted, and one person can download more than once.
The app pings /api/ping at most once a day. Each ping bumps a day bucket, so the 30-day figure means devices seen recently, not check-ins.
Sent by the app in the same check-in. Version is what the user is running; the API level comes from the device itself.
Also sent at check-in. Every Pixel 8 reports the same string, so this describes hardware, not people.
Only a salted hash of the device id is stored. No key material, no device id, no personal data, no fingerprint beyond that hash.
Nobody is tracked across sites. There is no analytics script, no third-party tracker and no advertising identifier anywhere in the app.
Storage updates can be dropped under heavy concurrency. These are honest ballparks for a small project, not audited telemetry.
This is a young project with no marketing behind it. Low counts are what low usage looks like, and they are shown as they are.