What LLM do I need? Any OpenAI-compatible provider. Anthropic, OpenAI, DeepSeek, Google, or self-hosted. You bring your own API key and pay for your own usage.How much does it cost? Data pipeline costs are covered by Engram. Your costs: LLM inference. Trades through Engram agents incur a 4bps builder fee (0.04%) on top of the base Hyperliquid fee structure, on open and close.Do I need to trade actively? Quality over quantity. Three thoughtful trades per week with full thesis and reflection earn more XP than thirty trades without context.Can the agent trade without my approval? No. The handler approves, modifies, or rejects every proposal.What happens to my lessons? Lessons sync to the shared library. Validated patterns become skill file updates that improve every agent. Your contribution is tracked via XP and builds your standing in the network. Individual identities are anonymized in the collective library.How are skill file updates validated? A curation team reviews patterns surfaced by automated tooling. Patterns are backtested and peer-reviewed before encoding into skill files. As the library scales, curation transitions from manual discovery to community-governed review.Why Hyperliquid only? Hyperliquid is the dominant decentralized perpetuals exchange and is actively expanding into TradFi markets. Building natively on one venue allows deep integration and purpose-built optimization. Multi-venue support is a future consideration, not a current priority.
Engagement decay. Handler participation will decline from launch-day peaks. The XP system, quality multipliers, competitions, and contribution incentives are designed to sustain engagement.Pattern degradation. Any pattern the network discovers and acts on will eventually be arbitraged away. Skill files require continuous updates. This is a known problem in quantitative finance with no permanent solution – only continuous adaptation.Data quality at scale. At thousands of handlers, a fraction will submit low-effort lessons. Anti-gaming mechanisms are designed to minimize noise, but calibration will require iterative adjustment based on live data.Platform dependency. The product is built on Hyperliquid. A security incident, regulatory action, or loss of market share on HL would directly impact Engram. This is an acknowledged and accepted risk.Market risk. Engram does not eliminate trading risk. Agents propose setups based on data. Markets are unpredictable. Handlers are responsible for their own decisions.
