You should not zero pad here because as is your frequency of 10 Hz falls directly on a DFT bin.
A =1;
t = 0:0.01:1-0.01;
x = A*cos(2*pi*10*t);
xdft = fft(x);
plot(abs(xdft))
Your data, x, is 100 samples in length. So you get 100*A/2 at -10 Hz and +10 Hz. That is equal to 50.
The zero pad length does not affect the amplitude, just the length of the data vector.
You can of course do:
xdft = fft(x)/length(x);
plot(abs(xdft))
