Here's my solution so far - create a variable that'll hold the generated types in and then change the code of the sentence list to never use any of these already-generated types:

// Create a new "takenTypes" array variable to store the already generated types
takenTypes = [[]]

sentence
  [takenTypes = [], ""]You find yourself in {a} [d=description.selectOne] [b=biome.selectOne].<br> As the area is [d.singularForm.lowerCase], [td = typesdescription.selectOne, uniqueReturn(td)] type 'mons are rather prolific. That is in addition to the [removeDups(others[b]).selectUnique(2).map(x => uniqueReturn(x)).join(" and ")] types that live in the [b] already. <br>Oddly enough, there also seem to be some [s=removeDups(stragglers).selectOne, uniqueReturn(s)] type 'mons in the area as well.

// Returns a new list that does not contain the entries already shown in the output
removeDups(list) => return list.selectAll.filter(x => !takenTypes.includes(x));
// Adds the entry into the "takenTypes" variable so they never show up again in the same output
uniqueReturn(i) => takenTypes.push(i); return i;