Count unique values from multiple items

[markfree]

A host has multiple items with a status string value of either "Ended OK" or "Ended Not OK".
I need to create another item that counts how many items have each value.

To find the value I am looking for, I was thinking of using the "find" function like this:

find(//ctm.job.status[*],,,"Ended Not OK")

With that I could try to use the "sum_foreach", but I'm not sure how to make them work together.
Something like this, but valid.

sum_foreach(find(//status[*],,"eq","Ended Not OK")

Any suggestions?

[markfree]

So far, I've found this reference for the "foreach" functions.

Conditions expression
The conditions expression supports:
operands:
group - host group
tag - item tag
"<text>" - string constant, with the \ escape character to escape " and \
case-sensitive string comparison operators: =, <>
logical operators: and, or, not
grouping with parentheses: ( )
Quotation of string constants is mandatory. Only case-sensitive full string comparison is supported.

I didn't understand how a "<text>" can be an operand.

If my expression is

count_foreach(//ctm.job.status[*]?["Ended OK"],5m)

When I test it, Zabbix gives the following error since aggregate calculations do not support strings.

unsupported calculated value result "double vector[0:0]" of type "double vector"

And when I try to add the item with the "count_foreach" expression, another error is displayed.

Invalid parameter "/1/params": incorrect expression starting from "count_foreach(//ctm.job.status[*]?["Ended OK"],5m)".

So, this "<text>" string constant as an operand doesn't make sense to me.
Also, I could not find a single example with just a string as an operand.

[markfree]

This all looks like a frequency distribution to me, and the status values are the actual classes, but I can't wrap my head around it.
Has anyone been able to work with categorical variables in Zabbix?