I looked through the rules and found no clear definition on how to handle agents regarding marks and OS score.
Agents are programs running on the deck but they are also personas. For the moment ignore the possibility of two decks.

It seems to me something similar to IC and hosts. When an IC marks a target all IC and the host itself share that mark.

1) After the errata, it is no longer clear if the contrary is truth: a mark put on the host also affects the ICs. Is there any official ruling on this?

2) I wonder if Agents and Deckers share their marks and if it works on both ways (receiving and putting marks).

3) Do agents accumulate a separate OS score or it accumulates it together with the Decker?

4) Can agents simultaneously operate on a different grid and host than the Decker? If so, it seems mandatory to keep a separate OS score to handle who will be attacked by GOD convergence. In addition, a positive answer here would imply that agents and deckers really use different matrix recognition keys (marks).

I think that question 4 is the key to all answers. If the agent must work on the same grid and host as the decker than it would share marks and OS score, similar to an IC. If the agent is free to go where it wants than it must have OS and mark of its own.

Well, the errata seemingly wanted to reverse the previous direction, not make it happen in both directions...


The basic rule is that marks are specific to one persona (and an agent is a separate persona). No rule without exceptions, but I can't find such an exception being mentioned for agents, so it looks like they indeed need separate marks by RAW.


I can't recall anything banning agents from using another grid. But as sprites show, GOD may even decide to smite the owner of an entity which has a separate OS, therefore the reverse conclusion (that a separate OS is needed to decide who gets smited) does not really hold. Agents may have their own OS or share OS with their owner, and we don't even know who gets hit by GOD once the agent's OS exceeds 40.

Moral of the Story: Don't use an Agent for anything illegal and you won't have to worry about generating OS, and therefore the Agent will remain beneath the notice of GOD.

nah, the moral of the story is that agents pretty much suck unless you have a separate device entirely for them to use, and even then they aren't exactly amazing.

Thank you guys for the answers! I just do not agree on one point. If the agent has its own matrix recognition key (does not share marks) than it must have his own OS score. It does not make sense otherwise.

Yes and no though. Ultimately the Agent is still just a program running off your deck and if it's OS count triggers a GOD response its going to shut down the deck end of things.

I thought moral of the story was: Don't Get Caught.

Why that? Is there any rule about marks which would not work if the agent shared its OS with the controlling decker?

Because have its own Matrix recognition key means that he is an independent user even if he shares the same origin point (deck).
I suppose it is like many users that are logged on out of a mainframe (host?). If one such user starts making illegal activities he would accumulate an OS for himself not all the host users. It also means that when GOD converges they will discover where he originally logged in.