Before updating this script, you need to find out if it's working well for you, rather than wasting your time and our time making a dead-end script. For example, your use of $istok() means that it only matches complete words, and is defeated by anything touching the badword, including punctuation like commas, or periods.
Also, the way you're doing this is looping through the variables N times when you have N swearwords, so this can bog down when you add more words. It would be even slower if you did this in a way which requires accessing a disk file to scan it for bad words.
Rather than using $read to check a swearwords.txt for a match, you can add bad words to a hashtable, then use $hfind to make a single 'pass' against $strip($1-) looking for a match. You can create a hashtable where the badwords are the item names, or are the itemnames themselves. You would also need to choose whether to add wildcard or regex patterns to the hashtable, because $hfind() matches them differently. i.e. * and . mean different things in wildcards than in regex.
If you haven't already done so, check out
https://www.mirc.com/help.html where more detailed syntax and examples have been added to some - but not all - identifiers and commands.
