This paper proposes a novel agent personality representation model used to provide interagent adaptation in modern games, coined as the Tactical Agent Personality (TAP). The TAP represents the tactical footprints of a game agent using a weighted network of actions. Directly using the action probabilities to model an agent's personality, removes the time and effort required by experts to craft the model as well as eliminates the performance dependency on expert knowledge. The effectiveness, versatility, generality, scalability, and robustness claims of the TAP architecture and its variations are applied and evaluated across a variety of game scenarios, namely, First-person shooters (FPSs), real-time strategy (RTS) games, and role-playing games (RPG), where they are shown to exhibit plausible adaptive behavior.