Isn't it? My understanding was that agents could be instructed to act autonomously. Basically, if you "set the agent free" on your commlink, with a blackhammer program, it would be the same as Black IC.

Calypso

Actually IC don't lauch themselves, the system/node launches them (top page 223) in response to an alert. Now i suppose you could say that a standard Agent is not directly launchable by the system alert. In that case you'd need an extra IP for a [cheap rating 1] bootstrap IC to spawn the higher rating Agent?

P.S. Does an independant Agent itself count in the program count compared against the System limit? Or do you just count the programs the Agent has loaded under itself?

There isn't a way to give them the keys to the node, so to speak? You definately can give them the keys to Databombs, right?

I'm not sure what kind of risk there would be in doing that? Which brings up another interesting topic. If an independant Agent doesn't recognize anyone as their 'owner' to get it to change load-outs does that mean they cannot be Spoofed directly like IC? Instead the malicious attacker would have to directly thwart a second firewall (as Agents seem to have firewalls) to get inside the independant Agent if they wanted to control it and turn it on the node?

Agents acting independently still have an owner. And yes, you can give Agents the code to a node so that IC doesn't attack it.

Calypso

Here be a can of worms - pryed open.

Basically, an Agent/IC loaded onto a Node/Persona counts a 1 Program for the Agent itself.
Then, all those Programs loaded onto it would count towards its own System, reducing it's Response.

So, basically, running an agent gives you free active programs - just, instead of using them, you have to order the Agent to do so, which it will, on it's own Initiative.

Funny sidenote:

Nodes are allowed to use Programs, too, substituting the Skill with their System, while replacing the System with a Pilot gives it more decision making capability.

Looks like all you need is a rigged up Commlink, serving a a Hacker-in-a-Box.

Ya, that's what i was thinking when i saw that. Which actually doesn't make much sense to me at all for something like a Commlink. For a large node, sure. They have a per session limit but have a large reserve of processing power in the entire node. But in a Commlink if there is that extra ability WTH wouldn't the system already be using that if it could? *shrug*

It's like the Matrix system design is missing a stat for nodes (including commlinks) that is a session/context limit. This is the number of personas (which an independant Agent, IC, or Pilot program also count as) that can run on the node before performance degradation.

More like a Hacker drone with the person sending the general commands via AR/VR akin to a rigger in the Captian's Chair.

Oh, I would certainly hope not. But, I see no reason to not have an agent toting around my mundane-yet-necessary programs, like Analyze, Browse, Edit, etc. If a really tough task needs to be done, that I don't think the agent can handle, I swap out programs temporarily.

Calypso

Yes, for generally all the reasons that a 'runner is ultimately better than a vehicle drone.

Or I let the Agent screw up, get attacked by the IC, and then slip through myself. There are no rules on Aggro, to use an MMO term, but if nothing else you draw out the defenses where you can see them. As the Active Alert is per user and long as no hacker is alerted, you're golden...

(Data Bombs hopefully excluded...)

Unfortunately that can bring a security hacker into the picture if there is one on call. If he isn't a total mellon head he will start looking intently for things amiss....like someone trying to run a diversion with an Agent. Also since tallies are per/persona, and thus the risk of detection increases in a reverse exponential way per extra invader, it is reasonable to assume are going to be more Hacker teams happening. So that should now become security hacker SOP for responding to an alert, trying to assess how many are on the inside so he knows if it is time to escalate to more drastic measures (node reboot, call for more backup, etc.)

the main danger would be a hostile decker spoofing your agent and forcing it to disarm the databomb. of course, he'd have to figure out (or guess) that the agent has the key. he might do it anyway, though, letting the agent take the hit from the databomb--i don't believe databombs reset immediately; if the agent sets off the databomb, there should be a short delay where the hostile decker can perform the necessary operations before the databomb resets.

Databombs don't reset. Once they crash, they're crashed until you reboot or reset it manually. (Or have an agent do it I suppose).

Ya, if the independant Agent can recieve instructions from the "owner" then that communication could be Spoofed. Wasabi seemed to be suggesting that kind of communication couldn't occur between the owner and independant Agent, which i didn't think was right but didn't feel confident to refute. *shrug*

That's a pretty crude technique, though effective when crude is good enough. The Databomb text does seem to suggest a one time only deal, and you can have only one attached Databomb per item. So at the least something would have to rerun the program to set another one back on (if the Databomb wasn't destructive to the item it was guarding).

P.S. I was thinking another really cool use of Spoofing the Agent/IC would for a one-shot stealth attack. It is my understanding that spoofing the Agent to attack it's owner wouldn't nessasarily reveal the spoofer, while directly attacking the owner is effectively an automatic revealing (barring some extraordinary conditons that don't come to mind right now). Even if the scan succeeds in revealing the Spoofer (assuming they haven't moved on already, or hadn't spoofed the command to the Agent in a different node), unless the victim's second Simple Action of that IP is spent countercommanding the Agent, the Agent will get to attack again if it gets an IP before the victims next.