We propose a formal language to represent and reason about signal transduction networks. The existing approaches such as ones based on Petri nets, and \pi-calculus fall short in many ways and our work suggests that an artificial Intelligence (AI) based approach may be well suited for many aspects. We apply a form of action language to represent and reason about NF/kappaB dependent signaling pathways. Our language supports several essential features of reasoning with signal transduction knowledge, such as: reasoning with partial (or incomplete) knowledge, and reasoning about triggered evolutions of the world and elaboration tolerance. Because of its growing important role in cellular functions, we select NF/kappaB dependent signaling to be our test bed. NF/kappaB is a central mediator of the immune response, and it can regulate stress responses, as well as cell death/survival in several cell types. While many extracellular signals may lead to the activation of NF/kappaB, few related pathways are elucidated. We study the tasks of representation of pathways, reasoning with pathways, explaining observations, and planning to alter the outcomes; and show that all of them can be well formulated in our framework. Thus our work shows that our AI based approach is a good candidate for feasible and practical representation of and reasoning about signal networks.