SentinelMind

Our journey began with a disturbing realization: while AI makes us more productive, it also makes attackers more dangerous by allowing them to craft the perfect psychological lie. We were inspired by a deep dive into the Big Five Personality Traits (OCEAN), where we learned that every successful social engineering attack is actually a precision-guided strike against a specific human vulnerability—whether it’s weaponizing Agreeableness through fake helpfulness or exploiting Neuroticism through manufactured urgency. We realized that if attackers are using psychology to break in, we must use psychology to lock the door.

To turn this vision into reality, we built SentinelMind. We developed a robust Chrome Extension using JavaScript to intercept threats directly in the browser, injecting "Psychological Speed Bumps" into the user interface. The "brain" of our project is a custom Machine Learning backend built in Python, trained on a specialized dataset of over 1,200 samples to identify five core attacker vectors: Urgency, Authority, Fear, Empathy, and Curiosity. We integrated Claude as our advanced linguistic engine to explain the manipulation tactics to the user in plain language, and utilized ElevenLabs to create a professional voiceover that brings our defensive vision to life. By profiling the attacker rather than the user, SentinelMind provides the defensive acceleration needed to outpace modern threats and effectively patch the human firewall.