So this sort of answers a question I had. I was thinking of pulling something similar with a few cheap decks and link them all to a primary deck. Then each one has an agent program running on it and I can essentially be a technomancer with some decent sprites assisting me. The drawback is that I would rule it that even if the agents are all assisting me that my OS score should spike faster by doing this. However, to be honest the way RAW is that an agent assisting doesn't generate OS because it isn't outright attacking it is lending it's support. So my question is if I link say 3 cheap decks (or even certain commlinks) together and using them to give me an assist. This is for a one shot big brute force attack not to be used all the time though I think it could be useful.
I don't think there's anything in the RAW that would prevent multiple agents from helping you, except that they each take a program slot on the device, and team work tests bonuses max out at whatever your rating in the skill are. And as far as the agents generating overwatch, if they're performing illegal actions, they would, but it would be attributed to their persona. Unfortunately their persona, if I understand it correctly is tied to your deck, so in the event of convergence, GOD would still know where you were.
I'm not so certain that people assisting in a teamwork test generate Overwatch. In a teamwork test, the people assisting are making their test unopposed, adding their hits to the leader's dice pool. The Overwatch increase is based on the number of hits that the target gets on the defense test. Now, it could very well be that since everyone is involved in the same action against a single defender, they would all share in the same amount of Overwatch gained, but it isn't really clear.
I took a look at Teamwork tests again on page 49, and it doesn't list a limit on the test... It very clearly indicates that when you make a teamwork test, you roll the "appropriate skill + attribute." The lack of it mentioning applying the limit might be an oversight, but it also could be an indicator that the limit wouldn't apply in this case (since the maximum bonus applicable is set by the leader's ranks), and the leader makes the final test normally with the limit determined normally (+1 for each assist).
Agents are Programs. You can't run more than one program of the same type on your deck. So you can't run multiple agent on your deck.
I'd argue that since RAW states that agents have their own persona this means that any two agent programs cannot be identical. Thus, each agent program is a unique program in and of itself, and the usual limitation of not being able to run two copies of the "same program" on a device doesn't apply.
Mostly, I feel this isn't a problem because all you're doing is increasing your attack surface. Since all agents would share the same condition monitor and have worse dice pools than most dedicated hackers, you'd be risking almost casual destruction for a few dice. But that's a personal opinion.
Kuirem makes a good point on this one, plus there is book rules to back it up. You can only load one agent at a time on a single device.
You can’t run more than one program of the same type on your deck at once (and no, changing the name of one copy of a program to run two copies doesn’t work, chummer).
An agent runs as a program and can use programs running on the same device as them.
