Because this has gone unanswered, and using this very nice answer as a reference, I can answer this part way. Source article here.
it was found that if the mass of second body becomes larger than
1/1200 of the first body, horseshoe orbits become unstable.
So Earth mass orbiting Jupiter (1/318) won't work according to the research above. You could try Mars mass objects orbiting Jupiter (1/2970) and you might get that to work for a fairly extended number of orbits. Horseshoe orbits are never stable over the long term the way trojan orbits can be, but they can be stable over numerous orbits if they are properly set up.
I tried to work out the specifics for higher mass horseshoe objects, but my attempts at calculations have just sent me in circles so far. If I make any progress I'll post an update. You'd have to move the Mars-mass moons quite far from the planet and you'd need to keep the variation in their semi-major axes quite small. Janus and Epimethius have only 50 km variation in their orbits, which is 1 part in 3000 of their orbital radius.
I'd explore variations in that range, and I'd start at the outer edge of stable region of Jupiter's hill sphere, and if you can't get it to work, decrease the mass some more, maybe lunar mass.