red-teamlisted

# Red Team The goal of this stage is **honest adversarial pressure, not reassurance.** Your job is to find real reasons this candidate should NOT be built, using only evidence already gathered — not to confirm it's a good idea. This is the last automated stage before `HUMAN_REVIEW`; whatever you don't surface here, the human reviewer won't see flagged for them. ## When to use this Opportunity is in state `DIFFERENTIATION_ANALYSIS` (after the `differentiation` skill completed). ## Steps 1. **Get the prompt.** ``` hunt redteam-prepare <opportunity_id> ``` Requires `DIFFERENTIATION_ANALYSIS` state. 2. **Work through the full checklist actively**, from `references/red-team.md` §2 — for each item, ask whether evidence already gathered (competitor autopsy, VOC findings, browser research, demand intelligence, economics) plausibly connects to it. Do not manufacture a risk with no evidence connection, and do not soften a well-evidenced negative finding to make the candidate look better: - Demand illusion - Seasonal illusion - Temporary OOS (out-of-stock) illusion - Review contamination - Suspicious review patterns - Margin illusion - PPC (advertising) dependence - Logistics problems - Supplier problems - IP/compliance risk - Dominant incumbents - Copycat vulnerability - Stale/expired data - Cross-source contradictions 3. **Cite `evidence_id`s for every `EVIDENCED`-labeled finding.** If a finding you believe