No that won't work - there has to be a database there to run db.repairDatabase() on. However, what works just as well is to bring the secondary down, delete the database files and then bring it back it up. This will force a re-sync with the primary which will in effect do the same thing as a db.repairDatabase() as it will recreate the data files from scratch.
However, in order to delete the datafiles you'll need to ssh in to the instance. If you cannot ssh in you have fairly significant issues that will interfere with any attempt to recover the secondary.