Human-agent teams exhibit emergent behavior at the team level, as a result of interactions between individuals within the team. This begs the question how to design artificial team members (agents) as adequate team players that contribute to the team processes advancing team performance, resilience and learning. This paper proposes the development of a library of Team Design Patterns as a way to make dynamic team behavior at the team and individual level more explicit. Team Design Patterns serve a dual purpose: (1) In the system development phase, designers can identify desirable team patterns for the creation of artificial team members. (2) During the operational phase, team design patterns can be used by artificial team members to drive and stimulate team development, and to adaptively mitigate problems that may arise. We describe a pattern language for specifying team design patterns, discuss their use, and illustrate the concept using representative human-agent teamwork applications.