No of context items fetched by cursor

Hey,
I have started using the cursor today and found its responses impressive. But, I have a doubt about its functionality. When I ask certain query to it from the repo in workspace, then it is fetching very large no. of items (I guess its near to 100 or more than that). Why is it always fetching those large no. of items for a single query. Even, sometimes I have seen that for a simpler query, whose answer is just a single line in 1 of the file of repo, it is fetching this large no. of items. So, is it any way to set a limit on no. of items fetched and make performance better?