Mann Kendall Test... which?

[Ginko75]

Dear trends earth developers,
I am new in GEE and I have very short experience.
I am trying to "reproduce" IN GEE what Trend's Earth application does calculating
the "land productivity" by using MODIS NDVI mean yearly values.
I am using the same dataset "MOD13Q1.061 Terra Vegetation Indices 16-Day Global 250m", the same legend and the same annual series to compare my output with that of Trend's Earth (degraded, improved or stable pixels..)
The final output is the significance of NDVI trends by applying the Mann Kendall Test.
Unfortunately, if I compare my final raster with that of Trend's Earth there are some differences (pixels are classified differently).
The MK test I am applying is this from GEE: https://developers.google.com/earth-engine/tutorials/community/nonparametric-trends
Someone knows if this is exactly the same code applied in Trend's Earth?
I suspect that the code is slightly different... hence the differences I find.
It is possible?
Thanks to anyone can help me!!