Knowledge acquisition is a crucial problem for the design of Intelligent Tutoring Systems (ITSs). To overcome this problem, authoring tools have been proposed. Over two dozen of authoring tools have been built since the earliest days of ITS, but each of them focuses on a particular kind of ITSs such as constraint-based tutors or model-tracing tutors. In the context of the AMBRE project, we are interested in ITSs teaching problem-solving methods. Such ITSs enable learners to acquire a specific method in problem-solving. Despite of the variety of existing authoring tools, these tools do not meet our needs either because approach adopted do not match to AMBRE principle or because they do not allow to represent all knowledge needed to design an AMBRE ITS. We propose AMBRE-KB, an authoring tool to help authors to elicit knowledge needed for the design of AMBRE ITSs. This tool supports the acquisition of knowledge to be taught, and the description of problems to be solved. We present the authoring process and illustrate it using French verb conjugation domain. A preliminary evaluation shows that AMBRE-KB is successful in producing domains models but more thorough evaluation is planned.