The Agent Skills Directory

[SAFE]: The skill implements a robust interface for the Hyperliquid DEX without any detected malicious patterns. It adheres to the stated purpose of financial trading.
[COMMAND_EXECUTION]: The skill provides tools to execute financial transactions, including placing orders (hl_order), setting leverage (hl_leverage), and performing deposits/withdrawals. These are authorized actions performed via a secure wallet service interaction.
[EXTERNAL_DOWNLOADS]: The client fetches real-time market data, order books, and funding rates from Hyperliquid's official API endpoints. These are well-known services and do not involve executing untrusted remote code.
[DATA_EXFILTRATION]: The skill accesses account-related information (balances, positions, fills) which is necessary for its trading functionality. This data is used within the agent's context to report status to the user and is not sent to unauthorized third parties.
[INDIRECT_PROMPT_INJECTION]: The skill processes structured JSON data from the exchange API (market info, trade results). While this constitutes an ingestion of external data, the risk is minimal as the exchange API does not return attacker-controlled natural language instructions that could subvert the agent's behavior.
Ingestion points: Market data, account state, and fill history fetched from the Hyperliquid API in client.py and exports.py.
Boundary markers: Not explicitly present, but data is handled as structured objects.
Capability inventory: Subprocess calls (via wallet service requests) for transfers and trades in client.py.
Sanitization: Inputs are coerced to specific types (float, int, bool) in tools.py before being processed.

hyperliquid