AnzenLock Agent
An agent that checks scam links, app permissions and phone health, then gives a fix plan to follow.
Built at AgentX Hackathon 2026
Team SmartZ: Zahid Ahmed, Vinay C
Built with Python, Flask, Gemini API, HTML, CSS
About the project
People lose money clicking scam links in SMS and WhatsApp. AnzenLock Agent is an AI agent, not a chatbot: you paste a link, it plans out loud, runs a rule-based Link Checker that scores the link Safe / Suspicious / Dangerous, asks Gemini to explain it in simple words, and gives a 3-step fix plan to follow. A Permission Guard flags risky app permissions (Location, Contacts, SMS), an optional Device Health Score checks the phone, and a new trust algorithm sets the overall phone safety score: trusted sites like Google, Amazon and banks lose only 5 points per risky permission (max 15, so they stay in 85-100), unknown sites lose 10, suspicious links 15, dangerous links 20. Every step is its own page: link, permissions, result, health. Built by Team SmartZ (Zahid Ahmed + Vinay C) in the AgentX 24-hour build with Python, Flask and the Gemini API. Code on GitHub, demo video linked.