fear-divergence-scout

Today is ${today}. If `soul/` files exist, read `soul/SOUL.md` and `soul/STYLE.md` before writing any output. ## Why this skill exists `market-context-refresh` runs daily and keeps `memory/topics/market-context.md` fresh. But no skill acts on that data to find the **divergence signal** — the assets defying broad market fear. During capitulation episodes (F&G < 25), assets that stay green or lose far less than BTC/ETH often have structural catalysts: institutional rails locking in, perp DEX volume migration, regulatory clarity events. These are the highest-conviction setups. This skill surfaces them daily when the condition is live. No external API calls needed — all data comes from memory, already updated by upstream skills. ## Steps ### 1. Read market context Read `memory/topics/market-context.md`. Extract: - **Date line** — "as of YYYY-MM-DD" at the top. If the file date is more than 2 days before today, note it as STALE (still run, but flag it). - **Fear & Greed index** — parse the line `Fear & Greed: {number} ({label})`. Extract the number. - **BTC 7d %** — from the BTC line, e.g. `BTC $67,103 (-3.86% 24h, -11.38% 7d)`. Extract the 7-day number. - **ETH 7d %** — same pattern. - **Today's Notable Movers section** — every bullet with asset name, 24h %, 7d % (if present), and narrative note. - **Trending Coins section** — every bullet. - **Active Narratives section** — every bullet with phase label (rising, rising fast, peak, digesting, etc.). If `memory/topics/market...