Do clips work for audio-only episodes?

No. Clips are cut from an episode's video, so an episode without video cannot be clipped.

When your agent searches for moments, any moment whose episode has no video to cut from is dropped rather than turned into a clip. The transcript can still be searched (the words are indexed either way) but there is no video to cut the clip from.

If you want your episodes to be clippable, publish them with video. Fountain extracts the audio from an uploaded MP4, so one video upload covers both formats.