Automatic assessment has become an important technique to reduce the grading load on teaching staff while providing an exhaustive evaluation environment for students. Several systems have evolved over the years providing sophisticated evaluation capabilities. However, fully automated assessment covers only a portion of the overall evaluation requirements in a typical Computer Science course. In this paper we present a tool for automatic assignment management that aims at satisfying several objectives. First, to support the large variety of assignment types and grading policies under the same paradigm by means of a phgeneric architecture. Second, to provide a phmulti-agent, scalable platform to cope with large enrollment classes. And third, to provide phfull connectivity with other administrative tools already present in educational institutions.