Risk disclosure
Risk & Custody Disclosure
HyperAgent is built for institutional desks, but automation and leverage carry material risks. Review these disclosures and confirm that your internal policies cover custody, onboarding, and crisis response.
1. Non-custodial architecture
HyperAgent never asks for or stores your private key. Signatures are produced either on your workstation (client signer) or on an external MPC/HSM. The Admin Console only records public metadata (public key, scopes, heartbeat), encrypted with KMS.
Because we cannot access private keys, you are responsible for keeping the signer online and rotating API scopes. If the signer goes offline, orders will fail until you restore the connection.
2. Algorithmic and model risk
LLM-driven strategies can behave unpredictably under novel market conditions. Past performance is not indicative of future results.
Models depend on upstream data feeds. Latency, exchange outages, or malformed data can cause poor decisions or missed trades.
3. Market and leverage risk
Trading perpetuals or other leveraged instruments amplifies both gains and losses. Sudden price swings may trigger liquidation before the brain has time to react.
Risk guardrails (VaR, drawdown, cooldowns) reduce but do not eliminate the chance of significant capital loss.
4. Counterparty and infrastructure risk
HyperAgent relies on the Hyperliquid chain and exchange APIs. Smart contract bugs, validator issues, or API downtime can interrupt execution.
System components (systemd services, PM2, networking) require correct configuration. Misconfigurations can halt bots or leave positions unmanaged.
5. Client responsibilities
Monitor wallets, balances, and signer health. HyperAgent cannot withdraw funds or top up collateral on your behalf.
Keep API scopes, environment variables, and signing devices secure. Compromised credentials may result in unauthorized trades.
Consult your compliance/legal team before operating in jurisdictions with regulatory restrictions.