hyperliquid
Warn
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [Indirect Prompt Injection] (MEDIUM): The skill ingests untrusted market data from an external API (https://api.hyperliquid.xyz/info) and processes it to format chat responses. \n
- Ingestion points: Data enters via the Info HTTP endpoint called in
scripts/hyperliquid_api.mjs.\n - Boundary markers: None identified in the provided documentation to distinguish API data from agent instructions.\n
- Capability inventory: The skill uses
nodeto execute scripts, performs network requests, and reads/writes to a local configuration file (~/.clawdbot/hyperliquid/config.json).\n - Sanitization: No sanitization or validation of the API response is mentioned before it is passed to the chat formatting logic.\n- [Data Exposure] (LOW): The skill accesses a local configuration file at
~/.clawdbot/hyperliquid/config.json. While this is used for account aliases, it establishes a pattern of local filesystem access which could be leveraged if the scripts are compromised.\n- [Command Execution] (LOW): The skill relies on executing local JavaScript files using thenoderuntime. This is a standard implementation but grants the skill the ability to run code on the host system.
Audit Metadata