To help software developers work efficiently, integrated development environments (IDE) include many tools. All too often, these developers are unaware of potentially useful tools within these IDEs that might help them complete their work more effectively. To improve both awareness and use of tools within an IDE, we have been developing a recommendation system called Spyglass that recommends tool(s) that might help a developer navigate information available in an IDE more efficiently. When designing such a recommendation system, important considerations are both the content of the recommendations and the form and manner in which those recommendations are made. In this paper, we focus on what we learned about the form and manner of making tool recommendations from a longitudinal user study of Spyglass. These results may be useful to others designing various kinds of recommendation systems for IDEs.