Guardian Angels: Using LLMs as personal security and productivity agents

A framework for deploying language models as always-on assistants that learn your patterns, intercept risky actions (like phishing clicks), and automate routine tasks. The approach treats LLMs as customizable agents rather than chat interfaces—useful for makers building personal productivity tools or exploring defensive AI use cases.