You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
CallShield is a spam call and text blocker for Android: fifteen-plus detection layers topped with a gradient-boosted tree model, a GitHub-hosted database currently at 32,617 spam numbers, real-time caller ID and RCS filtering. No API keys, no subscriptions, no accounts.
v1.7.28 is current, and the recent releases were about substance over motion: secondary text was brought up to WCAG AA contrast, the call log stopped re-aggregating itself on every update, and a batch of tests that could never fail got replaced with ones that actually exercise serialization round-trips.
On the language front, there's a complete Simplified Chinese translation in a community fork that I've invited upstream in issue #7, which also serves as the claim thread for any other language.
Detection quality is the whole product here, and it has two failure directions: spam that gets through, and legitimate calls that get flagged. The second one is worse, and both are invisible to me without reports. The thread pinned next to this one is for exactly that, no personal numbers required, patterns are enough.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
CallShield is a spam call and text blocker for Android: fifteen-plus detection layers topped with a gradient-boosted tree model, a GitHub-hosted database currently at 32,617 spam numbers, real-time caller ID and RCS filtering. No API keys, no subscriptions, no accounts.
v1.7.28 is current, and the recent releases were about substance over motion: secondary text was brought up to WCAG AA contrast, the call log stopped re-aggregating itself on every update, and a batch of tests that could never fail got replaced with ones that actually exercise serialization round-trips.
On the language front, there's a complete Simplified Chinese translation in a community fork that I've invited upstream in issue #7, which also serves as the claim thread for any other language.
Detection quality is the whole product here, and it has two failure directions: spam that gets through, and legitimate calls that get flagged. The second one is worse, and both are invisible to me without reports. The thread pinned next to this one is for exactly that, no personal numbers required, patterns are enough.
All reactions