Practical reasoning (PR), which is concerned with the generic question of what to do, is generally seen as a two steps process: (1) deliberation, in which an agent decides what state of affairs it wants to reach –that is, its desires; and (2) means-ends reasoning, in which the agent looks for plans for achieving these desires. A desire is justified if it holds in the current state of the world, and feasible if there is a plan for achieving it. The agent’s intentions are thus a consistent subset of desires that are both justified and feasible. This paper proposes the first argumentation system for PR that computes in one step the intentions of an agent, allowing thus to avoid the drawbacks of the existing systems. The proposed system is grounded on a recent work on constrained argumentation systems, and satisfies the rationality postulates identified in argumentation literature, namely the consistency and the completeness of the results.