A variety of syscalls aren't implemented yet, meaning that it cannot handle a few core libraries.
There is no parallel processing support.
Constraint solver backend isn't the greatest, only Z3 is really supported.
Angr cannot detect static libraries very well(though this probably isn't very important, since no one really uses them much)
Now we can work around most of these, and we most certainly want to use all the underlying projects that angr uses.
The question is, is using angr worth it, knowing that we will likely have to modify significant chunks of it to suit our purposes, or would it be simpler to directly use the API's of the underlying projects?
Angr is fantastic, but it has some limitations: