hyperliquid-advanced
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCREDENTIALS_UNSAFECOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the Hyperliquid API and WebSocket subscriptions.
- Ingestion points: Data enters the agent context through API responses and WebSocket feeds (e.g., 'twapStates', 'userTwapSliceFills') described in SKILL.md.
- Boundary markers: Absent; the instructions do not specify delimiters or warnings to ignore instructions within API responses.
- Capability inventory: Financial execution including order placement, automated cancellation scheduling, and API agent approval.
- Sanitization: Absent; no evidence of string sanitization for data returned from the exchange before it is processed by the agent.
- [CREDENTIALS_UNSAFE]: The skill describes processes that involve handling raw private keys within the agent's context.
- Evidence: The 'Approving an API wallet from code' section describes generating an 'agent_private_key' via the SDK, which is then returned to the agent's memory for use.
- [COMMAND_EXECUTION]: The skill executes high-impact financial transactions with real-world consequences.
- Evidence: Usage of 'exchange.order', 'exchange.schedule_cancel', and 'exchange.twapOrder' to manage assets and orders on the Hyperliquid platform.
Audit Metadata