Hookd is an independent research initiative focused on leveraging large language models to identify phishing attempts in everyday communications. Unlike commercial products, this self-contained project emphasizes privacy, using lightweight and transparent tools that operate directly on your device when feasible, avoiding centralized data collection or user sign-ups. Rooted in recent studies on explainable AI and user trust, Hookd aims to provide immediate, understandable feedback on suspicious messages. Recognizing the rapid evolution of phishing tactics, especially those powered by AI, Hookd combines offline models and optional server-side analysis to offer sophisticated yet private protection. This evolving tool encourages personal learning and user engagement in enhancing cybersecurity awareness.