I have a set of differentially expressed (DE) genes for each cluster calculated using Seurat. Now, I want to identify the enriched GO terms per cluster related to T-cell function. To do this, I extracted three lists (one per category) from the AMIGO database, each containing genes and associated GO terms specific to T-cell function. How can I use these three gene lists as a database in compareCluster with enrichGO, instead of using org.DB?

You might be able to do that, but in general I would doubt that the author of the package included functionality for a very specific edge case like that. The easy way would be to just use the correct OrgDb package and have enrichGO handle all the complexities for you, after which you could just filter to the three terms you care about.

Alternatively, if you want to 'roll yer own', do note that all any of these GO packages do is compute a hypergeometric test using phyper. If you have the set of genes in each term, it should be trivial to compute the required values for phyper - see the help page - and get the p-values for the three terms.