An advisor system is a system that proposes a just-in-time first-line help to the users of a given software, on the basis of an analysis of the actions and productions of the user. Within the Licef laboratory, two different types of advisor systems are developed: advisor systems dedicated to students using the Web-based interactive learning environments developed with Explor@ and advisor agents dedicated to courseware designers using the course design methodology MISA and its Internet-based workbench ADISA. The objective of the work presented in this paper is to elaborate a common methodological framework facilitating the design of these systems. The work presented here focuses on the identification of the pieces of advice; we propose an approach based on the analysis of a model of the considered task.