mob-check

Overview

Surface what real people are actually discussing, recommending, and arguing about on a topic right now, ranked by what they engaged with (upvotes, likes, views, comments, prediction-market money), not by SEO. Google aggregates editors; this searches people.

The skill is a fetch-rank-synthesize loop:

1. Parse intent and run a one-turn quality pre-flight (catch keyword traps before burning a search).
2. Fetch recent items from each available source using Hermes' native tools.
3. Pipe the gathered items through the bundled ranker (scripts/rank.py), which scores by relevance + recency + per-source-normalized engagement and fuses everything into one ranked, deduplicated, source-diversified list.
4. Read the ranked output and write a synthesized brief that obeys the output contract.