Launching tomorrow — would love your feedback on KidZoneSafe! 👋
by•
Hi everyone! 👋
After seeing videos of school bullying online — kids filming and posting it — I built KidZoneSafe: stealth live access to camera, mic, and screen, plus an "Intervene" mode for one-way video/audio with a siren to scare off offenders, no confirmation needed from the child.
It's built on peer-to-peer WebRTC (no AccessibilityService, no server-side storage), with my own TURN/STUN servers.
Launching here tomorrow — would love any feedback, questions, or thoughts before we go live! 🙏
24 views
Replies
@divvsaxena Yes — and here's exactly how it works today and what's coming.
Right now, a parent can silently connect to their child's phone camera and microphone during break time or after school. If they see or hear a threat — a bully, an aggressive situation — they activate Intervene Mode. This triggers a police siren alarm directly on the child's phone and/or opens a two-way voice session so the parent can speak out loud: "Step away from my kid." Tested it myself — unexpectedly effective. Especially the siren 😄
What's coming next: background scanning during parent-defined time windows. When a trigger is detected — crying, fighting, loud aggression — the system either sends an instant push notification to the parent, or first runs it through an AI speech recognizer for analysis (configurable in settings), then notifies. The parent then decides whether to connect and intervene.
So yes — from passive awareness to active real-time protection.
@divvsaxena The core — written by myself. Every AI told me it was impossible to bypass Android restrictions. Getting WebRTC running in the background was a nightmare — months of experiments. AI consultations were sometimes comical: ChatGPT and Grok kept going in circles arguing that Camera2 was outdated and CameraX was the way to go. Camera2 ended up working. Small detail, but telling about blind AI trust.
Android-Ionic sync including signal servers — myself. Screen capture — another story that took countless experiments to crack.
After that, especially with Claude Code, I started using AI more. Now I work as an architect. But you still have to control and write code yourself — AI can't hold the full picture.
@vitalii_polkhovskyi a feedback for you while using ai to write answers.
don't use --- em dashes.