Limitations in silkroad RL

[can312]

Hello guys, I need to choose a topic for my reinforcement learning class I thought maybe I could train 2 agents to PVP each other and learn optimum ways to use skills, what would be my limitations, it is hard for me to foresee as I'm inexperienced in both Silkroad developing and reinforcement learning.
If an expert can give me a guideline would be appreciated
Thank you for your help

[homelesshobo]

I'm not an expert, but making such an agent is definitely possible. Packets with the necessary information exist and are used in PvE bots for fights already.
A PVP bot would just have to additionally prioritize players and determine whether they are viable targets, using the pvp-flag (and changes to it) in the :

1   byte    FreePVP           //0 = None, 1 = Red, 2 = Gray, 3 = Blue, 4 = White, 5 = Gold

After that, it would just be a matter of the two opponents signaling combat readiness to each other (e.g. exchanging private messages to start or stop)
fighting to figure out the opponent's max HP, then prioritize the best skills to deplete it.


All of that aside, it would take a while to do that from scratch. You could look for bots and frameworks that work and add the pvp functions to them.
For example

1. The fundamentals for networking:
   Here's of using the security for logging in
2. Tutorials about development:
   The newer one based on rewriting sro code:
   The old, in depth one:
3. Information to write your own handlers:
   Here are two implementation examples: ,

[can312]

thank you, my biggest concern is making an emulator work on my computer and making it connect to my software so I can inject packets through. I will read the resources you sent Thank you again!