See what it keeps
Clotr's own page What Clotr stores shows every record it keeps, and what your browser allows it to reach: nothing outside the browser.
How it works, and your privacy
Clotr reads what you're about to send on your own device, and only there. Here's what happens, what it keeps, and how to check it yourself.
What happens when you type something private
- You type or paste as you always do: in an AI chat, an email, or a message.
- The engine reads it on your device. It looks for passwords, ID numbers, personal details and what scammers ask for, even when they're written out in words, like "five five five". Nothing is sent anywhere to do this.
- If it finds something, a note says what and lets you hide it. Hide it, or leave it in and send. It's your call: Clotr only stops a message if you've told it to.
One engine everywhere
Each part of Clotr runs the same engine, so a detail is caught the same way wherever you type: the extension today, and the apps for Windows and Android as they come. No AI inside: plain rules that anyone can read.
Your privacy
-
No servers, no analytics. Clotr doesn't send anything anywhere. A test checks that every single time the code changes.
-
It never saves what you typed. Just what kind of thing it found, on which site, when, and what you did about it. You can look at all of it and delete it whenever you want.
-
It only runs where you say. Out of the box, only on AI chats. On your email and chat apps, only on the ones you pick, and your browser asks you first.
-
No accounts. Nothing to sign up for, and nothing to sign in to.
Check it yourself
Watch it stay quiet
Open chrome://extensions, switch on Developer mode, and under Clotr click
service worker. Its Network tab stays empty while you use it.
Read the tests
A rule check fails every build that adds code able to send data anywhere.
Rebuild it
Every release can be rebuilt from the code, byte for byte: its checksum matches the release's
SHA256SUMS.txt.
Measured honestly
Tested, not just promised. I hid 6,200 personal details in 10,000 made-up messages. This build caught 5,027 of them, with 30 false alarms on the 3,800 that had nothing to catch. Messages typed on a phone are messier: there it caught 205 of 310. Real life is messier still, so if it misses something or warns about nothing, tell me.
A score that gets worse fails the tests.