Linux

If you know the .csv files are only buried, say, a max of 3 dirs deep then you can just do something like this:-

awk 'FNR==1,FNR==5 {printf("%s;%s\n",FILENAME,$0)}' *.csv */*.csv */*/*.csv */*/*/*.csv > output.csv

and that avoids a lot of the for/find fuckery.

(It'll complain to stderr if there are any directories that end in .csv but won't affect the output.)