yawning_titan.agents.fixed_red#
Classes
The FixedRegAgent provides the red activity for the FourNodeEnv specific environment. |
- class yawning_titan.agents.fixed_red.FixedRedAgent(skill=None, exploit_capability_dev=10, initial_no_of_zero_days=1)[source]#
The FixedRegAgent provides the red activity for the FourNodeEnv specific environment.
This agent has two linked concepts to make it more dynamic. The first is the concept of zero days. These guarantee compromise and work 100% of the time regardless of the vulnerability score. In addition, the agent also has the ability to develop additional zero days.
The behaviour of the agent is relatively fixed. The agent will prioritise the usage of zero days when available and otherwise randomly target nodes with a basic attack based on a configurable skill level.
- red_previous_node = None#
- red_current_node = None#
- select_action(uncompromised_nodes, compromised_nodes)[source]#
Select the Red Teams action for a time step.
- Parameters:
uncompromised_nodes – The list of uncompromised nodes linked to the Red team current position
compromised_nodes – The list of compromised nodes linked to the Red team current position
- Returns:
The action and target of the Red Team action
- update_location(target, red_current_node)[source]#
Update the current and previous location pointers.
- Parameters:
target – the node within the environment being attacked
red_current_node – the red agents current position
- do_red_action(red_action, skill_level, attack_sucess_threshold, machine_states, target, able_to_move=False)[source]#
Execute a red action and changes the environment state.
- Parameters:
red_action – The numeric value corresponding to the chosen action
skill_level – The red team skill level
attack_sucess_threshold – The attack power threshold for an attack to be succesful
machine_states – The current machine states
target – The machine being targetted by the attack
- Returns:
None