at this point, I feel like this thread, from my point of view at least, is more of a theoretical discussion.
I haven't done nearly enough research to say definitively if it's possible or not, so don't get your hopes up.
It's a project I may be interested in persuing, depending on API documentation, and what it actually has and can do.

that, in it'sself, isn't to me a major issue.  It should be easy enough to grab the data from objects and translate it to human speek.  My question is how can we grab a list of nearby objects to the controled character, sort them, and then move the mouse to an object.
That, to me at least, is the major hurdle.  Once you can get past that, you can do almost anything with some time.
build mode is also a major issue, how do I, as a blind person, build a house?  Representing that could be anywhere from very difficult to completely impossible depending on the level of api access we have.

@30, yes I have discord, user name is
annyminni
so your idea is  create a discord server for it?

I can understand that you folks are only having a discussion right now and still determining whether or not this can all even be done, so you take your time and figure it out. Trust me, I would much rather this be done right rather than have it rushed. I'll wait patiently in the meantime.

becompile all .pyc file sounds too much, but at least the wiki tells us the object  type , status, and how to change it by a bit coding, we need  to do is make it readable and output to the NVDA, like tolk or any lib, is it right? or i'm wrong?
if sighted people can create so much mod by change the object's status, i think we can read it? like if you need to change a person's hunger, you  first need to know the number, another mod can change allow teenager to do a little job, so you need to know how job was allowed and change the lower age, so you need to know the person's age as well.
in order to make the funtion we thought, letting teenager to do the job, we need to change a bit on these two parts