How to know if ID correspond to nucleotide sequence or protein with a script

If these are all GenBank or RefSeq accessions, you can use Entrez Direct for this as shown below:

$ cat accs.txt 
ATO98108.1
ATO98120.1
ATO98132.1
AVP78031.1
AVP78042.1
$ cat accs.txt | epost -db nuccore | efetch -format acc
## no output because none of them are nucleotide accessions
$ cat accs.txt | epost -db protein -format acc | efetch -format acc
AVP78042.1
AVP78031.1
ATO98132.1
ATO98120.1
ATO98108.1

NOTE: This will work only if the accessions are currently live because epost does not find any suppressed accessions. For example:

$ cat accs.txt 
NM_002826.3
NM_002826.4
NM_002826.5
$ cat accs.txt | epost -db nuccore -format acc | efetch -format acc
NM_002826.5

Here, all three accessions are valid nucleotide accessions but only the last one, NM_002826.5, is alive.

An alternate way is to use the accession prefixes defined here and come up with an appropriate regular expression query.