Stats at a glance. Four key metrics show pending, approved, rejected, and flagged items. Updates in real-time as actions are taken.
Filter tabs. All (default), Pending, Flagged, and Spam views. Active filter is highlighted with coral accent.
Queue items. Each shows title, author initials in colored circle, timestamp, flag reason badge, and sentiment dot (green/gray/red).
Bulk actions. Select multiple items (checkbox appears on hover), then bulk approve or reject. Floating action bar appears at bottom.
Split layout. Left side shows compact queue (360px fixed width), right side shows full detail panel with scrollable content.
Author context. Name, email, join date, and submission count help identify trusted users vs. potential spam accounts.
Spam score bar. Visual 0-100 score with color gradient (green→amber→red). Scores above 75 auto-flag for review.
Duplicate detection. AI-powered similarity matching shows related submissions with clickable links to those items.
Action buttons. Approve (green), Reject (red), Merge (purple), Request Changes (amber). Reject shows dropdown for reason.
Internal notes. Team-only textarea for leaving context for other moderators or flagging edge cases.
Spam threshold slider. 0-100 scale, currently 75. Items scoring above this are auto-flagged for review. Realtime value display next to label.
Auto-approve toggle. When ON, users with 5+ approved submissions bypass moderation queue. Threshold is configurable via slider below.
Blocked words textarea. Newline-separated list. Submissions containing these terms are auto-flagged or rejected depending on settings.
AI moderation toggle. When ON, Claude analyzes all submissions for spam, toxicity, off-topic content, and provides a confidence score.
Email notifications. Two toggles: instant alerts for flagged items, and daily digest at 9 AM with summary stats.
Toggle design. Coral accent when ON, subtle gray when OFF. Smooth animation via CSS transition on knob position.