What problem does it solve?
It turns public Hyperliquid (HL) data into labeled, verifiable perp-trade datasets and produces Reppo mint/vote intents without requiring ongoing manual intervention.
Core Features & Use Cases
- Builds labeled HL perp trade datasets from leaderboard-ranked wallets by joining HL user fills with HL OHLCV and deriving signal taxonomy, market context, verification hashes, and aggregate metrics.
- Enforces strict data freshness and rubric compliance by reading the current run’s
.hl-cache/ files (or degrading via web fallback) and skipping wallets/datasets that can’t be verified or don’t meet minimum evaluation volume.
- Writes autonomous intent files for datanet 9 (TradingGymAI) by hashing candidate datasets, selecting top performers by Sharpe (tie-broken by trade count), and emitting
.pending-reppo/mint-*.json plus .pending-reppo/vote-*.json where applicable.
Quick Start
Run the skill in your Aeon/Reppo pipeline and let it generate mint and vote intent files from the freshly pre-fetched .hl-cache/ inputs for datanet 9 (TradingGymAI).