Scans OHLCV candlestick data for common chart patterns like doji, hammer, engulfing, and head-and-shoulders formations, then classifies each by type, direction, and reliability score. Expects you've already ingested market data via market-ingest, pulls it from memory, runs pattern detection across single to multi-candle sequences, and stores results in either the typed pattern store or a namespace-routable memory slot. Honestly most useful if you're building a signals pipeline and need structured pattern data you can query later. The dual-path storage is a bit quirky but gives you routing flexibility depending on whether your downstream tools care about namespaces or typed records.
npx skills add https://github.com/ruvnet/ruflo --skill market-pattern