amazon-discoverylisted

# Amazon Discovery Runs the engine's existing zero-cost bulk discovery funnel (`hunt discover`) and reports its output faithfully. This skill drives one primary command — `hunt discover` — plus `hunt deepen-start` to hand off survivors to the next stage. It never re-filters, re-classifies, or re-scores a candidate itself: `hunt discover` already ran the deterministic cheap-eligibility filter and relevance classifier before printing anything. Full rule detail — funnel stage order, budgets, and exactly what RELEVANT/AMBIGUOUS/UNRESOLVED_TITLE/rejection mean — lives in: - `references/amazon-discovery.md` — funnel stages, config keys/defaults for every budget, stop-reason vocabulary, provenance/dedup rules. - `references/relevance-and-filtering.md` — the relevance classification algorithm and the cheap eligibility filters (ASIN shape, prohibited keywords, price band). Read both before interpreting `hunt discover` output for the first time. --- ## Operating procedure ### 1. Establish demand context (optional but preferred) If a demand signal was already recorded for this concept (`skills/demand-intelligence`), use `hunt demand-rank` / `hunt demand-show` to decide whether and how much to prioritize it. This is advisory only — proceeding straight to discovery without demand context is a valid choice, never blocked. ### 2. Choose the category deliberately `--category` must be one of the verified `CATEGORY_NAMES`. In `--mode real` it is **required** and must never be g