Fans of narrative works often want to recommend titles to others, especially in fandoms. Such recommendations, however, risk exposing listeners to spoilers, and prior countermeasures have largely treated spoiler handling as binary, either showing or hiding a statement—an approach that, when it hides content, silences the speaker's own message. We present UNSPOILER, a system that lets users write a review recommending any of several well-known titles, and instantly rewrites only the spoiler-containing passages into spoiler-free expressions, leaving the rest of their voice untouched. Users can see their own sentences parsed into a dependency tree and which words and characters triggered the rewriting. Also, they can compare their original review side by side with the unspoiled version—experiencing firsthand how spoiler avoidance and fandom's shared enjoyment can coexist.
