resolve_version truncated the plausible list to eight — a version ranked ninth was silently unreachable, distinguishable from absent only by API spelunking. Every plausible version now ships (the ballot sorts by score, so the tail costs nothing), and version cards gain "list every printing", wired to the existing open-versions endpoint, for when even the plausible set doesn't hold the right box — cues can mislead as well as shortlist. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_016jXZFSTZQKzAC8fqpWSz9g