Iv been messing around with mimicing human behaviour, like walking, clicking distribution from the target, even implemented situation based decision making. But hats amateur as i used some mouse clicks recorders and filmed real persons game sesions. 14 of my bots got banned after running 2months up to 23 hours a day average around 15-16h a day, but thats like less than 1/3rd of my army. Totally i am running this bot over 3 months now. Statistics arent very accurate but they are sick in a good way: Fastest ban was 100k resources collected. Few are at 250+k and running. Most are at around 50-80k range and running. Areas are suicidal, accounts fresh and all very low cmb lvl no other skills

Whole script antiban is misslicks/clicking patterns, decision making. No standart antiban apart in one place i use mousemovingobject.
While control group i gave/sold to chineese/lithuanians got like 80% banned before bond runned out. Control group used same walking methods but had no decision making and no dynamic antiban. Basicly antiban idea was clarity like assigning mood. for example semi afker, movie watcher. Then reducing overall reactions times and addint lazyness within the amount of time spent.
Clarity mud rcer has good antiban and it inspired me to add something complex. And it has like only few accounts banned. But in the future this wont be enough. Today 2 bans tomorrow 90% accounts.
Jagex is constantly improving their botwatch, we need to improve our antiban too. To counter that we need to gather data of missclicks from the target patterns. Maybe redo/update mouse movements. But for that we all should collaborate. We need to add antiban in srl include maybe in a functions deviations such as missclicks and better clicking patterns. Maybe add lazyness/response time/impatience/spammclicking into include functions too.