In a research community each research er knows only a small fraction of the vast number of tools offered in the continually changing environment of local computer networks. Since the on-line or off-line documentation for these tools poorly support people in finding the best tool for a given task, users prefer to ask colleagues. however, finding the right person to ask can be time consuming and asking questions can reveal incompetence. In this paper we present an architecture to a community sensitive help system which actively collects information about Unix tools by tapping into accounting information generated by the operating system and by interviewing users that are selected on the basis of collected information. The result is a help system that continually seeks to update itself, that contains information that is entirely based on the community’s perspective on tools, and that consequently grows with the community and its dynamic environments.