About Stop Biting

Built by a nail biter, for nail biters: Igor Gazivoda bit his nails for over 20 years, then built the awareness component of Habit Reversal Training into an app: MediaPipe hand tracking, 21 landmarks five times a second, running entirely on your own device.

The founder's story

I'm Igor Gazivoda, a software developer. I bit my nails for over 20 years: constantly, automatically, without noticing until the damage was already done. I tried everything: bitter polish, reminder bands, sheer willpower. They all failed the same way, because the habit is automatic and happens below the threshold of conscious awareness.

When I read the research on Habit Reversal Training, I understood why everything else had failed. HRT's core insight is that awareness is the bottleneck: you can't interrupt a habit you don't know is happening. I had a webcam, I knew how to code, and I knew MediaPipe could run hand tracking on-device. So I built the awareness component, the part of HRT that is hardest to do alone.

What Stop Biting does

Stop Biting uses your computer's webcam and Google's MediaPipe framework (compiled to WebAssembly and running entirely in your browser) to detect when your hand moves toward your mouth. When it does, an audible alarm fires. That alarm is the external awareness signal HRT research identifies as the most critical component of treatment. The app also logs each detection with a timestamp, so you can see your actual biting frequency, not your estimated frequency.

Privacy: the non-negotiable

All video processing runs in WebAssembly on your device. No video, no frames and no detections are transmitted. You can disconnect your internet connection after the app loads and it will function identically. Your alarm and bite log is stored in your browser. The site counts page visits with Google Analytics; that never includes anything from your camera.

The technology

Detection is built on Google MediaPipe's Hand Landmarker, which locates 21 hand landmarks. Stop Biting runs it five times a second: enough to catch a hand on its way to your mouth, and far less CPU than running it on every frame. Mouth proximity detection compares hand landmark coordinates to facial landmark coordinates in each frame. Read the full explanation at how it works.

How this site is written

Every article, guide and comparison here is written by me. I'm a developer, not a clinician, and the editorial policy and corrections page says exactly what that means for what you read: how claims are sourced, how anything I say about a competing product is checked, and how to tell me when something on this site is wrong.