NCaputoD function is based on Caputo L1 scheme or Caputo predictor-corrector scheme?

May I know if the NCaputoD in-build function in mathematica is based on Caputo L1 scheme or Caputo predictor-corrector scheme?

For ordinary users this information is hidden what NCaputoD methods uses,but using command Trace:

   Trace@NCaputoD[(2 Log[1 - x] (x + Log[1 - x]))/x^2, {x, 1/2}, -1]

  (*... NIntegrate...*)

we see a command NIntegrate by default in output.

The question is what method by default does NIntegrate use?

From Here we see:

NIntegrate uses symbolic preprocessing to resolve function symmetries,
expand piecewise functions into cases, and decompose regions specified
by inequalities into cells. With Method->Automatic, NIntegrate uses
“GaussKronrod” in one dimension, and “MultiDimensional” otherwise.

This web-link and this link maybe helps.

Regards M.I.