Hi, I understand gozz has mentioned that he prefers us not to look at the provided code for the other agents, and that we should devise our own strategy to solve this project. Are we allowed to use the variables in the agent.py class, namely the mission sizes, spy count and betrayals required dictionaries in our strategy?

Also are we allowed to use the fact that one of the agents is a random agent in our decision making? Not necessarily looking at the code in it, but just the fact that the actions of it are random, and so we can't use its actions to determine whether it is a resistance or a spy, but for the other agents we can

ANONYMOUS wrote: You are explicitly required to subclass the Agent class from `agent.py`. My previous statements have advised students not to analyse the benchmark agents provided (BasicAgent, SatisfactoryAgent), as you should be solving the problem yourself from scratch, and if we have reason to believe you have plagiarized from the benchmark agents you may receive zero marks. Agent and `agent.py` are not mentioned in this, as it is not an agent, but just the interface and rules information available to your agent.

ANONYMOUS wrote: Your agent will be tested (as in the tournament code) against various combinations of agents. It is not valid to assume that any particular agent (or any agent at all) will play randomly. Furthermore, if you were to do so, this would simply mean that you are optimizing your play against the random agent, and a better agent can be expected to play at least that well, as the better agent could always *choose* to play randomly. I encourage you to swap out what agents are in the tournament to assess the performance of your agent(s) against different combinations. Remember that it will only use whatever agents you have put in the agents directory. As you develop better agents, you can put them into rotation and see if you are able to develop one that consistently outperforms them.