Just launched TokenSwift-Pro. It compresses AI prompts using a TextRank-based
algorithm that runs entirely in the browser — nothing gets sent to a server.
Free tier gives 5 compressions/day. I built this mostly as a learning project, and
caught some real bugs along the way — including one version that actually made
prompts longer while claiming to shrink them, and a "100% preserved" badge that
wasn't telling the truth. Fixing those taught me a lot about being honest with what
a tool actually does.
Would love feedback, especially if anyone finds a prompt where it drops something
it shouldn't.