Argyris and Schön distinguish espoused theories - those which people speak about - from theory-in-use - those which can be inferred from action. In small software teams, developing reflective thinking about action is a vital necessity in coping with change. We address these issues in a Masters of Software Engineering, performed with an alternation between university and industry. University periods are dedicated to a long-term project performed in a reflective practicum. It aims to develop a repertoire of practices which helps young engineers deal with the 'messiness' of situations. Such a practicum provides students, working in groups, with the possibility of reflecting on action. We propose using the Course-of-Action framework to record observable aspects of the actor's activity into semantic wikis. Two hypotheses are discussed (1) self-analysis and self-assessment help to reveal theories-in-use; (2) the Course-of-Action observatory helps maintain awareness of the repertoire. A case study of a 6-apprentice team illustrates the observatory use and the reconstruction of apprentices' activity. Primary conclusions are that self-observation and self-analysis of a software engineer's activity help raise awareness of the initial structure of the repertoire. We are however unable to conclude that it helps reveal their theory-in-use (wh