An attack agent is the new name for anybody or machine that attacks a computer or network. These attacks can include viruses, spyware, disabling attacks on machines or networks, changing or stealing data, etc.
Terms that used to be used were hackers, crackers, cyber terrorists, etc. Hackers were referred to as black hat, gray hat and white hackers with white hat hackers being people who were trying to find weaknesses in computers and networks so they could be fixed. Black hat hackers were hackers who attacked computers or networks to do damage and gray hats cam in between.
The problem has developed that the terms were no longer clear what group attackers fitted in, they often belonged in several groups of type of attacker and when it was thought about all of them were doing some type attack. The terminology has changed to refer to everyone who does something that could damage a computer or network an attack agent.
The main problem I see with this generic term is that attack agents we used to identify as white hat hackers are all grouped in with the bad people. White hat hackers were quite often employed by companies to find security weaknesses in their computers, networks and software and many of them have no intent to every cross over on the bad side. It would be like saying undercover police are criminals since they may do illegal activities while undercover to catch criminals such as buying drugs.
Quite often in network attack people set up machines to do the attacks on the networks so in that case the person setting it up is an attack agent and the machine sending pout the code or messages to disable another machines is also an attack agent.