No problem. One time pad is known since 1915~1920: is very easy, fast and NSA proof.
Here is the heart of the program. There are two input files, the readable one and the key, and an output file: the coded one.

  OPEN "I", #1, FILEIN$:LFILEIN# = LOF(1)
  OPEN "I", #2, FILECHIAVE$:LFILECHIAVE# = LOF(2)

IF LFILEIN# < LFILECHIAVE# THEN CONTA# = LFILEIN# ELSE END

  OPEN "O", #3, FILEOUT$
   
     FOR NCARATTERE# = 1 TO CONTA#
       CARATTEREIN$ = INPUT$(1, #1)
       CARATTERECHIAVE$ = INPUT$(1, #2)
       NUMEROIN% = ASC(CARATTEREIN$)
       NUMEROCIFRA% = ASC(CARATTERECHIAVE$)
       NUMEROCIFRATO% = NUMEROIN% XOR NUMEROCIFRA%
       CARATTERECIFRATO$ = CHR$(NUMEROCIFRATO%)
       PRINT #3, CARATTERECIFRATO$;
     NEXT
     
  CLOSE

Decoding uses the same program, with the the coded file as input.
Coding/decodig statement is the one with XOR inside.
My problem is the coded file that may include EVERY byte from 0 to 255, I can "clean" the key but NOT the the coded file.

Tha attachment is a BINARY file with a short example, open with an Hex editor, NOT with a text editor.