Now, very few of the words are what one might call strict autoantonyms: which is to say, words which can have two opposite meanings in the same context. Sometimes, the difference in context is so clear that even calling the term autoantonymous seems generous:
strikeObviously, "strike" as in "miss" is a technical term; the latter is the common usage.
• to miss (e.g. in baseball)
• to hit; collide with
More often, the words appear at first blush to be strict autoantonyms; but I think that a careful imagining of the circumstances in which the word might be used will show two non-intersecting sets. For instance:
buckleThe sort of things one buckles are very different than things which buckle; and I don't think there's going to be a case where "seatbelt buckles" might mean that it came apart, nor one which "a building buckles" would mean it fastened. (At least none come to find. Can anyone else think of one?) Similarly "clip" (meaning fasten, as with a paperclip) and "clip" (meaning cut, as in fingernails) are simply used in such different contexts that it is much a pair of homonyms as a single autoantonym.
• to fasten
• to come undone; give way; collapse
I think these cases are the most common: a thing weathers well (meaning it persists) very differently than a thing is weathered (meaning worn); you bolt a door and bolt away, but it's only autoantonymic if you carefully define one to stick in place and the other to run away; otherwise they're just homonyms. No one would confuse something that's custom made (i.e. unique) with something that is customary (i.e. common); one's things, the other's habits. And so forth. Most are like that.
But there are others, and these cases slide, almost imperceptibly, into strict autoantonymism: the cases blur, the categories get less distinct, and soon you're getting words that are very nearly ambiguous: "it's fine", could mean just fine (i.e. ok), or very fine, i.e. high quality.
In practice, of course, we don't confuse them; the sets, while not strictly non-intersecting, and more carefully defined than it appears. As J. L. Austin wisely said, "our ordinary words are much subtler in their uses, and mark many more distinctions, than philosophers have realized".
But I think that "peer", "rent" and possible "temper" are pretty strictly autoantonyms.
Check out the full list and see what you think for yourself. If things seem strictly autoantonymous, however, be sure you're imagining very precisely the sorts of situations and cases in which the word is used; they may be "subtler in their uses" than you realize.
The same site, incidentally, has an entire page on -nym words. Many will be very familiar to educated readers. but speaking personally I'd never seen:
aptonymn (apt names, such as the famous Supreme Court case Loving v. Virginia)...and a number of others too. A number of them seem likely to have been coined but rarely if ever used -- we might term them dictionymns, words that exist only in dictionaries, perhaps? Or even coinonyms, words that were coined but never caught on and are rarely used save in reference to the coining. But, of course, all dictonyms and coinonyms can become real words if we choose. So take a look and see if any meet your fancy.
bacronym (a backwards coined acronymn, in which one starts with the final word and picks words to match)
capitonym ("a word which changes its meaning and pronunciation when capitalised; e.g. polish and Polish, august and August, concord and Concord.")
exonym ("a place name used by foreigners that differs from the name used by natives; e.g. Londres is the French exonym for London, Germany is an exonym because Germans call it Deutschland." -- That one's particularly useful.)