Logics for agents are useful when specifying, implementing and verifying agent programs. We show that modal logic provides a nice tool to define informational, motivational and dynamic aspects of agents. We conclude by showing how an agent programming language can also benefit from this modal approach.